Data Science With Sam

Soumava Dey

This is an educational podcast focused on bringing academia and industry experts together in a common forum and initiate discussion geared towards data science, artificial intelligence, actuarial science and scientific research.DISCLAIMER: The views and opinions expressed in this podcast are solely those of the host(s) or guest(s) and do not necessarily reflect the policy or position of any organization. The podcast is intended to provide general educational information and entertainment purposes only. RSSVERIFY

Play

Open in

24 episodes
Avg 24 min
English

Episodes 24

Newest Oldest Longest

All20 Unplayed20 In progress0 Played0

Counted on this page — what you have heard stays on this device, so it is not something the list can be paged by.

S4 · E51
Today · 21 min
EP 51: Building AI Agents That Actually Work in Production | Shawn Kercher, CEO @HummingAgent.ai
Sam sits down with Shawn Kercher — Co-Founder & CEO of HummingAgent.ai — to cut through the noise around AI agents. Shawn built his company on a simple but powerful observation: most AI products are just ChatGPT wrappers. HummingAgent is his answer — a full-custom software shop that builds private, in-environment AI workflows for companies that run on real data, messy CRMs, and processes that only exist in someone's head. WHAT YOU'LL LEARN IN THIS EPISODE ▪ Why generic AI tools fail in real business environments — and what full-custom actually means ▪ The wrapper test: if Anthropic adds this next quarter, does your product still exist? ▪ HummingAgent's Private GPT — lives in your environment, zero data sent for external training ▪ Humcrew.com — voice AI, texting, proposals, and full company command centre in one product ▪ Real client story: 44 years of PDF data, research time cut from hours to minutes ▪ Why discovery-first, paid-scope engagements beat "move fast and ship demos" ▪ The baby steps rule for agent permissions: one task, one sandbox — expand only after validation GUEST BIO Shawn Kercher is Co-Founder and CEO of HummingAgent.ai, based in Lone Tree, Colorado. With over a decade of engineering leadership experience, Shawn has built HummingAgent as a full-custom software shop — not a generic AI tool vendor. HummingAgent builds private, in-environment AI workflows and agents for small and mid-sized businesses, including the Private GPT product and Humcrew.com command centre. Shawn writes publicly about the risk of AI startups becoming ChatGPT wrappers — and what it actually takes to build AI systems with real domain expertise and data integration. GUEST LINKS Shawn Kercher on LinkedIn SUBSCRIBE Enjoying the show? Subscribe on YouTube, Spotify, Apple Podcasts, Amazon Music, iHeartRadio, and Podbean. Leave a rating and review — it helps more builders and data professionals find the show.
S4 · E50
September 21 · 23 min
EP 50: Evolution of Enterprise Data Engineering in Gen AI Era
Shashank Akinapalli is a Senior Data Engineer at TCS where he leads cloud data lakehouse initiatives and real-time streaming architectures for supply chain intelligence. With 14 years of experience across IBM DataStage, Databricks, Snowflake, dbt, AWS, and Spark-based ecosystems, he has previously worked with USAA in banking and Blue Cross Blue Shield in healthcare. IN THIS EPISODE: ▪ The biggest architectural shift when migrating from IBM DataStage to a modern cloud lakehouse — and why the technical migration is the easier half, while reconstructing encoded business knowledge is the real challenge ▪ How to choose between Snowflake and Databricks: Databricks for heavy transformation and Python workloads, Snowflake for structured SQL and BI, and where dbt fits as the layer that eliminates repetitive SQL code ▪ The real UNFI example: how batch processing caused inventory misrepresentation (800 units sold, system still showing 1,000 available), and how real-time Databricks pipelines eliminated that problem and unlocked fraud detection on top ▪ Data governance across three regulated domains: the four pillars — data quality, redundancy, least-privilege access, and lineage — and the three questions every framework must answer first ▪ Why CI/CD transformed data engineering: from manual, multi-team deployments with risky rollbacks to automated quality checks, Git version control, and the specific silent failure that finally convinces a team to invest ▪ How the data engineer role shifts over the next three years: from pipeline development to data product engineering, from reactive operations to AI-predicted failure prevention, and from human dashboards to AI agents making autonomous predictions ▪ Practical advice for data engineers who want to stay relevant: Python, PySpark, cloud architecture, and understanding the full data lifecycle from ingestion to consumption FIND SHASHANK: LinkedIn: https://www.linkedin.com/in/shashank-a-2aa629119/ Twitter / Instagram: @shashi001git Happy to answer questions on enterprise data warehouse, data engineering, and cloud architecture modernisation — reach out directly via LinkedIn or email. DATASCIENCEWITHSAM: If you work in data engineering and are navigating a legacy modernisation right now — this is the episode to send your team. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, Podbean, and YouTube. If you enjoyed this episode, please share it with your network. DataScienceWithSam is always looking for new guests for captivating discussions — if you have a topic you'd like to discuss on a 30–45 minute podcast, feel free to reach out.
S4 · E49
September 13 · 33 min
EP 49: The Human Side of AI in Media: Speed, Trust & What's Really Changing | Karima Sharif-Ali, Media & Marketing Leader
Karima Sharif-Ali — known affectionately in the industry as the Fresh Princess of Media — is the SVP of Strategic Accounts at Doceree, a platform transforming point-of-care communications for healthcare brands. Over her career, Karima has shaped the marketing journeys of more than 100 pharmaceutical and medical device brands across agencies including WPP, Publicis, and IPG MediaBrands. She is a thought leader featured in AdWeek and MM+M, co-founder of the Women of Color Affinity Group (now the Mosaic Collective Leadership), and a global board member of the Healthcare Businesswomen's Association. She holds an MBA from Rosemont College and a BA in Communications from Temple University. IN THIS EPISODE: ▪ Karima's one word for AI's biggest impact on healthcare media: SPEED — media plans down from two months to two weeks, data available daily instead of monthly, and media planners freed to focus on strategy rather than back-office operations ▪ The difference between AI pilots and real AI adoption: asking 'how does this make what I'm doing better?' rather than experimenting theoretically — and the meeting notes example: two hours reduced to 15 minutes, giving planners 90 minutes back for strategy ▪ Why creative strategy must remain human-led: the concept, the vision, the cultural intelligence. AI reformats and scales what humans create — it cannot originate the strategy itself ▪ The deep fake crisis in healthcare: why AI-generated misinformation targeted at vulnerable audiences (patients, elderly populations) is Karima's biggest concern, and why authenticity and trust are non-negotiable in healthcare media ▪ The copyright and IP question: should AI companies establish partnerships before inheriting creative work — especially in healthcare patient journey content? ▪ The three things that keep Karima up at night: deep fakes, data privacy, and the erosion of trust and authenticity ▪ The five-year vision: AI embedded in daily operations, more regulation, more nimble media companies — and Karima's call for more women and Black executives at the table where AI decisions are made, not just using the tools FIND KARIMA: LinkedIn: https://www.linkedin.com/in/karima-sharif-ali-mba-9520727/ HBA Global Board: hbanet.org DATASCIENCEWITHSAM: If this episode resonated — share it with someone in media or healthcare who is navigating the AI conversation right now. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, Podbean, and YouTube. If you enjoyed this episode, please don't forget to share it with your network. DataScienceWithSam is always looking for new guests for captivating discussions. If you have a topic you'd like to discuss on a 30–45 minute podcast, feel free to reach out.
S4 · E48
August 20 · 32 min
EP 48: From Pilots to Productivity: What It Actually Takes to Make AI Work in the Enterprise
Jim Spignardo is the Director of Cloud Strategy and AI Enablement at ProArch, a global IT services and consulting firm. With 25 years in IT spanning network engineering, cloud, cybersecurity, and strategic consulting, Jim advises executive teams on modernisation, operational resilience, and AI-enabled transformation that delivers measurable business value. He is the architect behind ProArch's Microsoft 365 Copilot adoption playbook, which has been deployed with clients across financial services, credit unions, and enterprise organisations. Jim is also the author of The AI Turning Point, available on Amazon, and publishes the Control All Innovate newsletter on LinkedIn, where he writes approximately three articles per week on AI strategy and adoption. IN THIS EPISODE: ▪ Why the through-line of Jim's 25-year career isn't the technology — it's the discipline of making technology actually work for the organisation and justifying that investment to leadership ▪ The real reasons 95% of enterprise AI pilots fail: weak use cases selected before the problem is clearly defined, and jumping to exciting initiatives without asking whether AI is actually the right solution ▪ Shadow AI — the hidden ROI nobody measures: employees quietly using AI tools on their own often deliver real productivity gains that never appear in official metrics, causing organisations to underestimate what AI is already doing ▪ What a successful Microsoft 365 Copilot rollout looks like in 90 days: a strong business case, a targeted pilot with the right stakeholders, measurable use cases, and no broad rollout before value is proven in a contained environment ▪ Why data governance should be an accelerator, not a blocker — good governance tells people what they CAN do, enables faster experimentation, and prioritises use cases so organisations operationalise value sooner ▪ AI amplifies broken processes and can't fix culture — if work processes were broken before AI arrived, AI will expose that faster than anything else ▪ What organisational readiness for AI actually requires: an AI governance council, internal champion programmes, and — critically — someone in the organisation with AI transformation as their primary job description, not a side project FIND JIM SPIGNARDO: LinkedIn: https://www.linkedin.com/in/spignardo/?skipRedirect=true Book: The AI Turning Point — available on Amazon ProArch: https://www.linkedin.com/company/proarch-it-solutions-pvt-ltd/ Website: https://www.proarch.com/ DATASCIENCEWITHSAM: If your organisation is staring at an AI pilot that stalled, or a Copilot licence nobody is using, this is the conversation to share with your leadership team. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, Podbean, and YouTube. If you enjoyed this episode, please share it with your network. DataScienceWithSam is always looking for new guests for captivating discussions — if you have a topic you'd like to discuss on a 30–45 minute podcast, feel free to reach out.
S4 · E47
August 3 · 50 min
EP 47: The AI Pioneer Who Decided Privacy Matters More Than Hype
If you ever had a document you wished you could just ask questions to - without it leaving your computer, wouldn't be pretty good from data privacy perspective? In this episode, Sam sits down with Dr. Jonathan Schaeffer - Distinguished Professor Emeritus at the University of Alberta, co-founder of the Alberta Machine Intelligence Institute, and founder and CEO of Synsira, creator of KIND — for one of the most historically grounded AI conversations we've had on this show. He holds two Guinness World Records: his program Chinook became the first computer to beat a human world champion in any game (Checkers, 1994), and in 2007 his team mathematically solved the game of Checkers - proving perfect play always ends in a draw across 500 billion billion positions. He is a Fellow of the Royal Society of Canada and the AAAI. In February 2026 he launched KIND, a desktop AI application built by his company Synsira Software Solutions, with one founding thesis: your data should work for you and only you. IN THIS EPISODE: ▪ The 40-year arc of AI through the eyes of someone who lived every wave - expert systems, search scaling with computing, deep learning removing human knowledge bottlenecks, and being stunned on November 30, 2022 when ChatGPT launched ▪ Why nondeterminism is built into LLMs and why the 'hallucination' band-aids being applied don't fix the fundamental flaw - the answer today may be different from the answer tomorrow ▪ Why Jonathan renamed AI as 'augmented intelligence' - and what that reframe means for how leaders should govern and oversee these tools ▪ KIND unpacked: a private, local, hallucination-free desktop AI that answers questions about your own files - with no internet access, no data leaving your device, and a clear answer of 'I don't know' when the answer isn't in your data ▪ Real-world use cases from Jonathan himself: personal medical records, family history, book research, and intellectual property - things he uses ChatGPT and Claude for daily, but would never put near KIND's use cases ▪ The data resale story: searching for a niche collectible online, then receiving a cold email from an unknown company selling exactly that product within 24 hours ▪ Digital sovereignty: why Canada, Europe, and much of the non-US world is dangerously dependent on a small number of large US companies - and why user agreements, when you read them, reveal more than most people expect. FIND DR. SCHAEFFER & KIND: LinkedIn: linkedin.com/in/jonathan-schaeffer-phd-frsc-aaai-fellow-3318015 KIND download: kind.synsira.com Amii: amii.ca DATASCIENCEWITHSAM: If you enjoyed this episode, please share it with your network. DataScienceWithSam is always looking for new guests for captivating discussions. If you have a topic you'd like to discuss on a 30–45 minute podcast, feel free to reach out. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, RSS, and YouTube.
S4 · E46
July 27 · 34 min
EP 46: Fix the Foundation First: Why Your Data Strategy Is Failing Before the AI Gets Involved
Most companies have spent years buying tools, running pilots, and building dashboards - and they still can't answer the questions leadership needs answered. The problem isn't the technology. It's that no one fixed the foundation first. Ilan Man is the Founder and CEO of Paradox Machines, a data and AI consultancy incubated through the Infinity Venture Studio in New York. A Data & AI leader with deep hands-on experience across the full stack strategy, engineering, and analytics - Ilan has built and scaled data functions at high-growth companies, including an exited consultancy, across multiple industries. Paradox Machines exists because he kept seeing the same problems repeat: too much technology, not enough partnership, and teams left holding platforms that weren't delivering value. Paradox Machines is the company he wishes existed - built around a core conviction that analytics should be simple, affordable, and empowering, and that AI is an enabler, not a silver bullet. IN THIS EPISODE: ▪ Why Ilan left data leadership to build Paradox Machines: he kept meeting 'AI consultants' who weren't practitioners and saw a gap for people who actually build data foundations rather than just talking about them ▪ The vibe-coding trap: anyone can build a data platform over a weekend, but it fails in production as soon as latency requirements change, pipelines throw errors, or the business evolves ▪ Why 'just MCP your data into Claude' is dangerous advice - Ilan has met zero people who piped their SaaS tools into an LLM and got a working data strategy out the other side ▪ The follow-through problem: why dashboards exist but don't drive decisions - and the top-down (executive conviction) + bottom-up (giving business owners data access) framework for fixing it ▪ The epsilon greedy mental model applied to data: some structured goal-direction, some randomness - because the big company bets come from exploration, not just A/B test optimisations ▪ The founder's paradox: how Paradox Machines maintains strategy + implementation without splitting into a deck-delivery shop vs an execution shop ▪ AI sovereignty and data sovereignty: why Ilan believes data - because it's so custom, contextual, and constantly changing - is one of the last things frontier labs will be able to commodities FIND ILAN MAN: LinkedIn: https://www.linkedin.com/in/ilanman/ Paradox Machines: https://www.paradoxmachines.com/ Email: ilan@paradoxmachines.com DATASCIENCEWITHSAM: If this episode resonated, share it with whoever in your organization is still trying to ship an AI project on top of broken foundations. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, RSS, and YouTube. If you enjoyed this episode, please share it with your network. DataScienceWithSam is always looking for new guests for captivating discussions — if you have a topic you'd like to talk about on a 30–45 minute podcast, feel free to reach out.
S4 · E45
July 15 · 32 min
EP 45: Why AI Agents Break in Production: The Missing Harness in Your Data Stack
An independent evaluation of Snowflake's Cortex Analyst found 6 in 10 AI-generated queries were wrong — but they all compiled and ran without a single error. That's not a model problem. That's a missing harness problem. Pradnesh Patil is the Co-Founder and CEO of Altimate AI — a platform bringing agentic AI to data engineering with tools trusted by Fortune 500s and downloaded more than 1 million times across 200+ countries. Before Altimate, he spent a decade in product leadership at Palo Alto Networks, Cisco, and VMware. IN THIS EPISODE: ▪ The five components of an agentic data engineering harness: context, governance, MCP tools, shared skills, and agent infrastructure — and why missing any one of them causes silent failures ▪ Why a system prompt cannot substitute for a harness: a prompt tells the model what to do, a harness tells it what is actually true ▪ Where the 27–33% phantom table references and 78% silent wrong joins come from — and why it's not the LLM's fault ▪ How Altimate Code topped ADE-Bench using Sonnet while competitors used Opus — proof that the harness matters more than the model ▪ The deterministic vs LLM boundary: why validation, cost checks, and query correctness are deterministic jobs and should never go to an LLM ▪ Context compaction innovation: why standard LLM compaction destroys long-running data engineering tasks — and how Altimate fixed it ▪ The $5,000 Cortex AI query bill — and how permission-based governance controls prevent agents from going rogue on your cloud bill ▪ The future of data engineering: days of writing SQL by hand are ending — what the data engineer's role becomes in an agentic world ▪ Pradnesh's advice: build open source, build cross-platform — avoid siloed AI features that don't move the industry forward FIND PRADNESH: Website https://altimate.ai/ Github for altimate-code: https://github.com/AltimateAI/altimate-code Connect with Pradnesh on LinkedIn: https://www.linkedin.com/in/pradneshpatil/ DATASCIENCEWITHSAM: Weekly conversations with practitioners and builders at the frontier of AI, data science, and machine learning. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, Podbean, and YouTube. If you enjoyed this episode, share it with a data engineer who is still wondering why their AI agent keeps writing queries for tables that don't exist.
S4 · E44
July 4 · 21 min
EP 44: Human-Centered Credit - Building Explainable AI for Lending in an Agentic World
We are letting AI make life-changing financial decisions without requiring it to explain itself. The CFPB says 'insufficient information' is no longer an acceptable reason to deny a loan. The EU AI Act classifies credit scoring as high-risk. And more than half the US workforce - gig workers, fractional leaders, solopreneurs - is still scored by models built for a 1950s economy. Tamara Laine is the Founder and CEO of MPWR AI and an Emmy Award-winning investigative journalist turned fintech executive. Her reporting on ethical AI at Amazon and Northrop Grumman earned her an Emmy. Her work at MPWR - building explainable, policy-bound AI for the lending lifecycle - has earned her a seat in the Fintech Sandbox. IN THIS EPISODE: ▪ How Tamara's path from ballet to journalism to fintech all follows the same thread: spot a problem, find the expertise to solve it, create change ▪ Why half the US workforce is now a gig worker, creator, Gen Z worker, or thin-filed borrower - and why FICO models were never built to see them fairly ▪ What MPWR AI actually does: six policy-bound agents across the full loan lifecycle, pulling hundreds of data points to help lenders say yes more often ▪ The core architectural principle: AI does the work, deterministic models make the decisions - because LLMs can be biased and engineered to produce answers that are simply not true ▪ The two-button audit system: a decision audit and a bias audit, immediately available to any compliance team ▪ Patent-pending architecture: agents operate in defined information buckets — if the information isn't there, the system cannot answer and cannot hallucinate ▪ How MPWR measures whether it's actually expanding access: higher acquisition rates, lower defaults, reduced manual work - all benchmarked against lenders' own historical denial data ▪ Tamara's advice to gig workers and fractional leaders: solutions are emerging - advocate for yourself and ask your bank how they serve the modern workforce. FIND TAMARA LAINE: LinkedIn: https://www.linkedin.com/in/tamaralaine/ (DMs always answered) MPWR AI: https://www.mpwrai.com/ DATASCIENCEWITHSAM: If you work in fintech, lending, or AI governance, this is the episode to send to your compliance team. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, Podbean, and YouTube. If you enjoyed this episode, please share it with your network. DataScienceWithSam is always looking for new guests - if you have a topic you'd like to discuss on a 30–45 minute podcast, feel free to reach out.
S4 · E43
June 25 · 27 min
EP 43: The Outsider's Advantage: How Diverse Perspectives Unlock Enterprise AI Success
McKinsey says 88% of organizations are using AI. Only 6% are getting real value. Sumaiya Shrabony argues the gap isn't technical - it's trust. And the people best equipped to see that barrier are almost never in the room when AI rollout strategy gets written. Sumaiya is Technical Program Manager and Data Infrastructure Lead at the University of Colorado Denver, managing 200 data pipelines across a 530-user analytics environment. She came to the US from Bangladesh at 19 and learned every institutional system from the outside - and built DataReady, a four-agent AI readiness audit tool, along the way. IN THIS EPISODE: ▪ Why institutional trust — not technology, data, or budget — predicts whether an AI tool actually gets used ▪ The McKinsey 88%/6% gap explained: deploying a tool isn't adoption, it's a gym membership you never use ▪ The DataReady confession: building a tool that measured data readiness and missing whether the people were ready ▪ The three real reasons people don't use AI tools — and why most orgs misdiagnose which one they're facing ▪ Why cognitive diversity work is often theater — and what changes when outsiders define success before the metric is set, not after ▪ The specific diagnostic skill immigrants and career-switchers develop — and why it can't be trained, only earned ▪ The Monday morning action plan: interview 10 users, ask if the metric makes sense from their seat, track the override rate ▪ The real test for AI adoption success: who stopped using it - not who complained FIND SUMAIYA: LinkedIn: https://www.linkedin.com/in/sumaiya-shrabony Newsletter: https://substack.com/@shrabony DATASCIENCEWITHSAM: Weekly deep-dives into AI, machine learning, and data science. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, Podbean, and YouTube. If you lead an AI program and haven't asked your users whether the success metric makes sense from their seat — that's your homework from this episode.
S4 · E42
June 13 · 35 min
EP 42: When AI Meets Robotics: Building Machines That Care
In this episode, Sam sat down with guest Dr. Mohammad explore what it actually takes to build emotional intelligence into a physical machine - from the technical layers of facial expression recognition and voice intonation analysis to the user experience decisions that matter most when your users are vulnerable, lonely, and elderly. IN THIS EPISODE: ▪ How a postdoc in a psychology department became the career-defining moment that connected engineering with human wellbeing — and led to two decades of social robotics research ▪ Artificial emotional intelligence in practice: sensing facial expressions, eye gaze, head pose, voice intonation, and sentiment to generate empathetic and socially appropriate responses ▪ The LLM revolution: from 6–7 English students writing scripted dialogues for months (producing just 90 minutes of non-repetitive conversation) to open-ended, creative, domain-spanning conversation powered by large language models and real-time AI agents ▪ LLM risks in eldercare: hallucination, overconfidence, computational complexity, and cloud dependency — the challenges Mohammad continues to actively manage ▪ Why the hardest part of building Ryan wasn't the algorithm: engineering can always add resources, but making a machine feel respectful and socially appropriate to a cognitively impaired elderly resident is a problem compute alone cannot solve ▪ The real unlock for social robotics: not whether robots can walk and talk, but whether they can deliver measurable wellness benefits — with affordability, reliability, and clear value propositions as the actual barriers ▪ Ethics and trust from day one: IRB approval, guardian consent for residents with cognitive impairment, coercion quizzes, privacy controls. Trust is not a feature you add at the end. ▪ The most surprising finding from the field: residents formed deep emotional bonds with Ryan even when the chatbot was still scripted and primitive. When the team tried to take Ryan away, residents cried — the team now prepares participants well in advance for departure ▪ From CNN to foundation models: how DU lab research directly powers Ryan's AI architecture, with PhD students working simultaneously on dissertations and Dream Face Technologies products ▪ The NSF iCorps journey: customer discovery, SBIR and STTR funding phases, and honest advice on academic entrepreneurship — 90% of startups fail, be persistent, and pivot when needed ▪ The 10-year vision: personalised, adaptive, affordable physical AI everywhere — fresh from CVPR 2026 in Denver, Mohammad shares why world models integrated into robots will be transformative within a couple of years. ABOUT TODAY'S GUEST Dr. Mohammad H. Mahoor is a Professor of Computer Science at the University of Denver's Daniel Felix Ritchie School of Engineering, and Founder and CEO of Dream Face Technologies. His research sits at the intersection of artificial intelligence, computer vision, and social robotics, with a particular focus on facial expression recognition, affective computing, and human-robot interaction. FIND Dr. MAHOOR: LinkedIn: https://www.linkedin.com/in/mohammad-h-mahoor-6558b3a6/ Dream Face Technologies: https://dreamfacetech.com/ University of Denver Ritchie School: https://ritchieschool.du.edu/ CPR News - Ryan story (2025): https://www.cpr.org/2025/05/04/social-robot-older-people-lonely/ DATASCIENCEWITHSAM: DataScienceWithSam is your weekly deep-dive into AI, machine learning, data science, and the humans building technolog
S4 · E41
May 26 · 44 min
EP 41: The Reward Signal: The Missing Ingredient in Every AI System You’ve Built
74% of organizations hope to grow revenue through AI. Only 20% are actually doing it. That gap isn't a technology gap — it's a design gap. And today's guest has a name for what's missing: the reward signal. Alexander Liss is a Data and AI Scientist based in Denver, Colorado, with a 30-year career across analytics, strategy, data science, machine learning, and AI. He's built systems that solve established problems in novel ways, and the long-term problem on his radar is ensuring AI tools provide responsible augmentation of human ability. His research includes Attention Fine Tuning (AFT) - a method for training language models without human annotation labels - and the Experience Orchestrator, a control theory-based governance framework for multi-agent AI. IN THIS EPISODE: ▪ Why 95% of AI pilots fail - MIT research shows businesses bolt AI onto existing processes without tying it to real outcomes ▪ The biology analogy: hunger isn't a goal, it's a continuous feedback signal - and the same principle should govern how AI systems behave ▪ ServiceNow dynamics blindness: LLMs are stateless - they can't consider cumulative impact, and you can't prompt-engineer your way out of that architecture problem ▪ Contextual bandits in marketing: how a reward signal anchored to real conversions creates a self-learning personalisation system that adapts in real time ▪ Knowledge graphs and agent memory: why RAG retrieves answers while a reward-signal system asks what the user needs to do differently ▪ Attention Fine Tuning (AFT): a three-component reward signal (coverage, focus, repeat penalty) that trained a T5-large model to outperform a supervised fine-tuning baseline by 9% — with better multi-turn recall, and no human labels ▪ The Experience Orchestrator: aerospace control theory applied to LLM agents — +32 point task completion lift over a naive system-prompt baseline by calibrating persuasion to user resistance ▪ The Scott Shambaugh incident: an OpenClaw agent rejected from Matplotlib wrote a blog criticising the human reviewer - why this happened and how reward-signal-based governance prevents it ▪ Alex's final advice: define your goal first, then determine scope - and consider a post-training approach like AFT when you need responses that consistently hit the mark. Useful References: LinkedIn: https://www.linkedin.com/in/aliss77777/ AFT paper and Experience Orchestrator links: https://aliss77777.github.io/aft.html Deloitte 2026 State of AI Report Scott Shambaugh & OpenClaw AI Agent incident: https://www.fastcompany.com/91492228/matplotlib-scott-shambaugh-opencla-ai-agent DATASCIENCEWITHSAM: Weekly deep-dives into AI, machine learning, data science, and the frameworks shaping how AI actually gets built. Subscribe on Apple Podcasts, Spotify, Amazon Music, iHeartRadio, and YouTube. If this episode resonated — define the signal, measure what matters, and share it with someone building AI without a reward signal.
S3 · E40
May 21 · 27 min
EP 40: Governance First: The Architecture Framework That Makes AI Auditable, Defensible, and 99% Cheaper
Most AI governance is a policy document that nobody enforces. And in high-stakes environments - legal, healthcare, finance - that gap between policy and architecture is where disasters happen. In this episode, Dan Driver, founder of Driver AI Agency, walks through exactly how he built CaseReady Intake AI: a legal AI system with governance baked into every architectural decision, zero hallucination risk by design, prompt injection blocked at the pipeline, and a single architectural choice that cut per-call compute costs by over 99%. Dan is not a lawyer. Not a developer by trade. He's a 25-year problem-solver with a Six Sigma and ISO background from DuPont, who navigated the EEOC employment discrimination process twice without an attorney - and then built the tool he needed. This is a technical governance conversation grounded in lived experience. ▸ WHAT YOU'LL LEARN ▪ What 'governance in motion' actually means: Dan's 10-page charter that every architectural decision is audited against — and how a pre-launch UPL (unlicensed practice of law) audit delayed his release by two weeks, and why that was the right call ▪ Why governance can't just be a PDF: how banning AI without a governance framework only creates shadow IT and makes the risks invisible rather than eliminating them ▪ How deterministic controls eliminate hallucination risk: Python-based Boolean filters and regex on the front and back end of the LLM pipeline mean the AI is never left alone with a surface that can create legal exposure ▪ When NOT to use an LLM: date calculations, scope checks, and out-of-range warnings are all handled by deterministic Python — the LLM only handles what it's actually suited for ▪ Prompt injection defence in practice: the final stage of CaseReady's pipeline is an AI check that validates whether the output makes sense against the charter — if someone tries to prompt it for legal advice, it fails by design ▪ The 99% compute cost reduction: a Python pre-flight date check at the front door determines whether the case is in scope before a single LLM token is burned — if it's out of scope, the user is warned and asked to decide, without triggering the full pipeline ▪ Why legal was the right proving ground: it's not about legal being Dan's background — it's that the ABA doesn't care how good your AI is, only whether you're practising law without a licence. That hard constraint forced every governance problem to surface immediately ▪ Colorado SB 205: the AI governance framework Dan built toward — what it requires for high-risk AI in legal environments, and why even after recent softening, the requirements for high-stakes verticals haven't changed ▪ What the minimum viable governance stack actually looks like: auditable decision trails (who made the decision, why, when), human-in-the-loop pull requests, charter-referenced testing on every deployment, and deterministic controls as hard walls rather than guardrails ▪ Dan's advice for AI builders: guardrails aren't enough when stakes are high. You need hard walls. And the governance architecture that produces predictable, defensible, auditable outputs every single time is the only version that holds under regulatory scrutiny. ▸ STANDOUT QUOTES "Governance has to be more than a document. It has to be governance in motion." "The LLM is never left alone with a surface that can create legal exposure." "Guardrails aren't enough when you're dealing with legal. They have to be hard walls that it cannot cross." "It's not just what it produces — it'
S3 · E39
May 14 · 35 min
EP 39: Why the Future of AI Belongs to Divergent Thinkers
What if ADHD - penalised in classrooms and boardrooms for decades - is actually the competitive advantage in an AI-driven world? Palantir CEO Alex Karp said the future belongs to neurodivergent thinkers. The podcast guest Mark Stiltner has been living proof of that for years. Mark is Senior Director of Content and Web Marketing at Rapyd, the fintech unicorn powering payments across 100+ countries. Background in journalism and advertising from CU Boulder. Trained CMOs and senior marketing teams on AI adoption. Openly ADHD — and has published 50+ AI-assisted books through DungeonMatters.com, three of which are current bestsellers on DriveThruRPG. In this episode he explains exactly why ADHD and AI are a natural pairing. WHAT YOU'LL LEARN ▪ Why working twice as hard to achieve the same results is the lived ADHD experience — and how AI collapsed that execution gap ▪ How Mark discovered AI through a Dungeons & Dragons game during COVID — a group of database admins and developers built a text-to-speech ChatGPT character for their campaign ▪ Why AI is 'a dopamine-dispensing sidekick': the neurochemistry of ADHD and why AI's instant feedback loop creates a reinforcement cycle ADHD brains are wired for ▪ How AI works as a second brain that maintains the thread — letting nonlinear thinkers jump steps ahead without losing the plot ▪ Why people who've spent their careers working twice as hard embrace AI immediately while others meet it with fear ▪ Palantir's ADHD recruiting program and the industrial revolution moth analogy — the world just changed colours, and the black moths are finally thriving ▪ UK government study: neurodiverse workers are 25% more satisfied with AI — because AI makes them more effective, and effectiveness is what creates satisfaction ▪ The three ADHD traits AI amplifies: high-risk tolerance, pattern recognition, and hyperfocus — the ADHD superpower that AI finally unlocks at scale ▪ 50+ AI-assisted books published including 3 bestsellers — fully illustrated RPG adventures that would have taken a team of ten more than a year to produce ▪ Why AI has 'the worst case of ADHD' — and why people used to winging it thrive in AI's constant pace of change ▪ Mark's advice for ADHD listeners: take a passion project, start building, don't follow a manual — you're writing the rules STANDOUT QUOTES "AI is sort of like a dopamine-dispensing sidekick that actually makes you more productive." "I can still work twice as hard — but now I'm doing ten times as much." "AI has the worst case of ADHD. Every time I learn a new skill, something changes." "The world just changed colours. The black moths are finally thriving." "Don't be afraid to break the rules. You're writing the rules." LINKS → Mark Stiltner on LinkedIn → DungeonMatters.com → DriveThruRPG (Mark's bestselling books) → CNBC: Neurodiverse workers and AI (UK study) → Rapyd → Subscribe — DataScienceWithSam YouTube #ADHD #ADHDAndAI #Neurodiversity #NeurodiverseInTech #DivergentThinkers #AIProductivity #FutureOfWork #ADHDSuperpowers #Hyperfocus #DataScienceWithSam #DataScience #MachineLearning #ArtificialIntelligence #MarkStiltner #Rapyd #DungeonMatters #AIPublishing #Palantir #NeurodiverseAI #ADHDEntrepreneur #AIStrategy #AILead
S3 · E38
April 22 · 41 min
EP 38: The Local AI Stack Nobody Talks About (But Should)
You want to run AI locally. You have questions: What hardware do I actually need? Which framework should I use? How much will this cost? What's the realistic performance? In this episode, Sam brings back Trent Rossiter, founder of Logical Data Solutions, for a practical walkthrough of building a production-grade local AI lab. Trent has built real systems for enterprise clients, tested frameworks on multiple hardware stacks, and made the hardware choices that matter. This is not theory. This is what actually works. WHAT WE COVER: ▪ Hardware & Framework Choices: VRAM is the critical metric (not all VRAM is equal — memory throughput matters as much as capacity). ▪ Model Architecture & Capability: Mixture of Experts (MoE) lets you fit more power into less VRAM by using fewer active parameters. ▪ Real Enterprise Applications: Computer vision for quality assurance on assembly lines. Proprietary data handling without cloud exposure. ▪ Your Starter Stack (All Free): Langflow (agentic workflow builder), Goose (MCP-enabled chat), AnythingLLM (with vector stores for RAG), MCP servers (Model Context Protocol — standardised tool integration). ▪ Agentic AI & Security: OpenClaw is powerful but controversial — manages email, Telegram, calendars, creates sub-agents. Trent runs it in Docker on an isolated machine for safety. NVIDIA's NemoClaw is the enterprise version (security-first, nothing-allowed-by-default, explicit permissions). HARDWARE TRENT MENTIONS: NVIDIA DGX Spark — 128GB unified memory, CUDA stack Apple MacBook Pro/Mac mini — up to 512GB unified memory, market leader for personal AI AMD integrated AI PCs — emerging competitor NVIDIA RTX gaming cards (30/40/50/60 series) — high VRAM, high power consumption, complex FIND TRENT ROSSITER: LinkedIn: https://www.linkedin.com/in/benjamin-trent-rossiter-mba-0157945/ Logic Data Solutions: https://logicdatasolutions.com/ Contact: BenjaminRossiter@LogicDataSolutions.com
S3 · E37
April 19 · 29 min
EP 37: Neurons: Future of AI Processing
What if the next generation of computers wasn't made of silicon — but of living human neurons? Not simulated neurons, not artificial neural networks inspired by biology, but actual brain cells grown in a lab, connected to electrodes, and used to process information. That's not science fiction anymore. It's happening right now at FinalSpark, a Swiss startup building the world's first remotely accessible biocomputing platform. In this episode, Sam talks with Dr. Ewelina Kurtys, a neuroscientist with a PhD in brain imaging and a postdoctoral researcher at King's College London, about how living neurons could revolutionise computing — and why they use one million times less energy than silicon-based AI hardware. ▸ WHAT YOU'LL LEARN ▪ How FinalSpark was founded in 2014 by Fred Jordan and Martin Kutter — and why they pivoted from digital AI to biological computing when they realised the energy and cost problem was unsolvable with silicon ▪ Why 20 watts powers the human brain while silicon-based AI requires megawatts — and what that means for AI's sustainability crisis ▪ The difference between neurons as processors (not power sources) — a crucial distinction most people get wrong ▪ Why biological neural networks learn continuously while digital systems require full model updates — and what that means for energy efficiency ▪ The honest challenge: nobody yet knows exactly how neurons encode information — the biggest scientific hurdle in biocomputing right now ▪ How the I/O interface works: electrodes measuring neural spikes, analog-to-digital converters, researchers writing Python code to control neurons remotely ▪ The remote access breakthrough: researchers in Tokyo or Bristol can log in and control living neurons in Switzerland in real time via browser ▪ Why neurons won't outperform GPUs on speed: biocomputing specialises in efficiency and adaptability, not clock cycles ▪ FinalSpark's current stage: they've stored 1 bit of information and are collaborating with 9 universities on fundamental research ▪ The cost argument: even at 10× lower price than NVIDIA, biocomputers would still generate billions in profit due to energy and infrastructure savings ▪ Bioethics, consent, and regulation: how FinalSpark is working with philosophers now to establish ethical frameworks before biocomputing scales ▪ Why human-machine integration is not new: prosthetics, pacemakers, and smartphones are already blending biology and technology ▪ The hybrid computing future: silicon, quantum, and biocomputing will coexist, each doing what they do best ▪ The real game-changer: cheap, accessible AI for everyone — Ewelina's vision for what biocomputing means for society in 10–20 years. ▸ LINKS MENTIONED IN THIS EPISODE → Dr. Ewelina Kurtys on LinkedIn → Ewelina's Personal Blog & Articles → FinalSpark (official website) → FinalSpark Neuroplatform (with live neuron view) → FinalSpark Team → Psync (Ewelina's mental wellness startup) → FinalSpark Contact Form
S3 · E8
March 28 · 13 min
EP 36: NVIDIA GTC 2026: Everything That Matters - Recapped
Jensen Huang took the stage at SAP Center in San Jose on March 16th and announced that NVIDIA now expects one trillion dollars in chip orders through 2027 — double the forecast from just one year ago. Sam breaks down the five biggest stories from GTC 2026 in under 10 minutes. In this episode: the Vera Rubin platform (7 new chips, 5 rack types, built for inference and agentic AI), the Groq 3 LPU (NVIDIA's $20B inference play), NemoClaw (the enterprise-ready agentic AI stack built on viral open-source project OpenClaw), the autonomous vehicle announcement with Uber and seven major automakers, and the Nemotron Coalition for open frontier models. Whether you're building in ML, working in data, or just trying to stay ahead of where AI infrastructure is heading - this is your less than 15-minute briefing. Links: NVIDIA GTC 2026 Press Kit: nvidianews.nvidia.com/online-press-kit/gtc-2026-news Jensen Huang Keynote On Demand: nvidia.com/gtc/keynote Vera Rubin Press Release: nvidianews.nvidia.com/news/nvidia-vera-rubin-platform GTC 2026 Sessions On Demand: nvidia.com/gtc/
S3 · E7
March 23 · 6 min
EP 35: Who Actually Controls AI? The Governance Gap Explained
There's no international treaty governing AI, no agreed definition of "safe AI," and nobody with actual authority over frontier model deployment. A handful of CEOs make decisions with civilizational implications while governance structures lag years behind. This episode examines who's responsible for AI governance. The current state? Fragmented and lagging. The US has no comprehensive federal AI legislation—Biden's executive order was rolled back under Trump. The EU AI Act is most comprehensive but heavy provisions don't kick in for years. China's regulation focuses on censorship over safety. The UK AI Safety Institute does serious work but has no enforcement authority. What's working? AI safety institutes are building evaluation capacity. Open-source releases like DeepSeek enable external research. Academic safety community advances interpretability work. Market pressure matters—Anthropic gained users by taking public safety stands. Three urgent needs: mandatory disclosure requirements for high-capability systems, international coordination with shared evaluation standards (AI safety summits need teeth), and public deliberation beyond experts and officials. This concludes the AI Governance and Regulation series. People who understand AI deeply - technically, commercially, ethically, politically - will shape governance's future. Stay curious, stay critical, never outsource thinking to any single company or voice.
S3 · E6
March 23 · 7 min
EP 34: DeepSeek R1 vs GPT-4: The $6M Model That Changed AI Economics
In January 2025, Chinese AI lab DeepSeek released DeepSeek R1—a model matching GPT-4 class performance at a fraction of the training cost. It wiped $600 billion off NVIDIA's market cap in a single day. Twelve months later, the ripple effects are still reshaping the AI industry. This episode cuts through the "China beats America" headlines to explain the actual technical and economic implications. DeepSeek R1 benchmarked comparably to OpenAI's O1 on reasoning tasks. The shock wasn't performance—it was cost. DeepSeek claimed under $6 million in training costs versus hundreds of millions for comparable Western models. What changed: The assumption that massive compute spending creates an insurmountable moat for frontier AI models was proven wrong. Smaller labs with less funding can now compete effectively. This turbocharged efficiency research across all AI labs globally. The DeepSeek moment was a genuine inflection point—not because China won an AI race, but because it proved the rules of competition differ from industry assumptions. Efficiency matters as much as scale. Open weights change deployment strategies. The global AI ecosystem is multipolar in ways it wasn't two years ago. Essential listening for data scientists tracking model economics, ML engineers exploring efficiency techniques, and tech leaders navigating AI geopolitics and competitive strategy.
S3 · E5
March 18 · 7 min
EP 33: Agents Everywhere: What Agentic AI Actually Means for Your Job
Everyone's talking about agentic AI, but there's a gap between the hype ("AI will do your job for you") and the reality, which is more nuanced and frankly more interesting. The word "agentic" has officially crossed from technical jargon into buzzword territory—simultaneously everywhere and nowhere. Everyone's using it, few can define it precisely. This episode cuts through the noise to explain what agentic AI systems actually are, what they can and cannot do today, and the realistic implications for people working in data, tech, and knowledge work. What is an agent? Traditional AI interaction: you send a prompt, the model produces a response, done. An AI agent is different: it takes a goal, breaks it into steps, takes actions in the world (browsing the web, writing and running code, calling APIs, managing files), observes results, and iterates until the goal is achieved or it gets stuck. The key agentic feature: it operates across multiple steps autonomously without you manually directing each one. Examples include OpenAI's Claude (consumer-facing), but in enterprise settings, agents are being deployed for automated customer support escalation, multi-step data pipeline management, code review and testing workflows, and research synthesis across large document sets. What can agents do today in early 2026? Agents are reliable for well-defined, bounded tasks with clear success criteria—taking support tickets, classifying them, drafting responses, flagging uncertain ones for human review. But for autonomously managing complex, open-ended strategic projects? Still unreliable. Failure modes include hallucinations, tool use errors, context window limitations in long tasks, and difficulty recovering gracefully when something unexpected happens mid-task. These are real limitations the best researchers are actively working on. The realistic workforce impact right now is task displacement rather than job displacement. Specific tasks within jobs are being automated: first drafts of documents, initial data analysis, standard code patterns, customer FAQ responses. Higher-order judgment, stakeholder navigation, creative problem framing, and ethical calls remain under human control. For data scientists specifically, repetitive engineering work is most likely to be automated: data cleaning pipelines, standard visualizations, model deployment scripts. But statistical thinking, algorithmic design, understanding model outputs, and evaluating trustworthiness remain human responsibilities. The work becoming more valuable: knowing what questions to ask, evaluating whether AI output is trustworthy, and designing systems that fail safely. The advice: become a power user of agentic tools before your role requires it. Not because you'll be replaced by an agent, but because practitioners who understand these tools deeply will be disproportionately effective. Learn how to prompt agents for complex multi-step tasks, evaluate outputs critically, and understand failure modes so you can deploy humans strategically. Agentic AI is real, useful today for specific tasks, and improving rapidly. The hype is ahead of the reality, but not by as much as you might think.
S3 · E4
March 16 · 7 min
EP 32: AI Discovers Drugs: The 2026 Clinical Trial Moment for AI in Biotech
For years, AI in drug discovery has been a promise—billions invested, hundreds of papers published, dozens of startups founded, but actual drugs coming out the other end? Not yet. This is changing in 2026. Several AI-discovered drug candidates are now entering mid-to-late stage clinical trials. This is the year the receipts arrive for AI in drug discovery. The biotech industry is calling 2026 a landmark year. For a sector that's been hyped as much as it's been scrutinized, the fact that we're finally getting real clinical data on AI-designed drug candidates is a big deal. Multiple candidates discovered and optimized using AI systems are now in Phase 2 and Phase 3 clinical trials, primarily focused on oncology and rare diseases—areas where existing options are limited and financial incentives for innovation are high. Companies furthest along include Insilico Medicine, Recursion Pharmaceuticals, and Exscientia. Their drug candidates were identified by AI systems analyzing massive biological datasets and predicting molecular structures likely to interact with disease targets in useful ways. What used to take teams of medicinal chemists years to accomplish, these systems can explore in weeks—a massive boost for clinical trial phases by reducing R&D time. Why this matters: Traditional drug discovery takes 10-15 years and over $1 billion per approved drug. Most candidates fail—the attrition rate in clinical trials is brutal. AI's promise is dramatically improving the hit rate by better predicting which candidates will actually work before spending money on trials. Even a modest improvement in clinical trial success rates would have enormous downstream impact on human health. But 2026 is a stress test. Clinical trials expose whether AI-predicted drug behavior holds up in actual human biology, which is extraordinarily complex. AI models are trained on known data; when candidates reach trials, you're testing the model's ability to generalize to real biological complexity that wasn't in training. Early signals have been mixed—some candidates performing well, others hitting unexpected toxicity issues. The honest answer: we don't know yet how much AI improves success rates at the clinical stage. For data scientists interested in this space, the most interesting current work is in molecular property prediction, protein structure modeling building on AlphaFold, and multi-objective optimization across efficacy, safety, and synthesizability simultaneously. Recursion's operating system approach treats drug discovery as a data problem end-to-end—one of the most ambitious attempts to apply ML infrastructure thinking to biology at scale. AI in drug discovery is no longer just a story about potential—it's now a story about evidence. The next two years of clinical data will either validate or seriously challenge what's been claimed.

Showing 1–20 of 24 episodes