DataFramed

The role of the machine learning engineer is being rewritten in real time. AI coding assistants are absorbing parts of the day-to-day, planning and evaluation are eating up more of the week, and the lines between machine learning engineer, AI engineer, and data scientist are blurrier than ever. For anyone working in data and AI — or trying to break in — this shift changes what skills are worth investing in, what employers actually screen for, and how interviews are run. What's still worth learning? What does a competitive portfolio look like? And how do you stand out when a thousand applicants are using bots to apply?
Marina Wyss is a Senior Applied Scientist at Twitch (an Amazon company), where she builds production AI and machine learning systems across content understanding, recommendations, and forecasting. She came into the field from a non-traditional background — a political science undergrad and a Master's in social data science in Berlin — and has held machine learning roles at Coursera and a Berlin-based statistical consultancy along the way. Outside her day job, Marina runs a popular AI/ML YouTube channel and weekly newsletter, and coaches people transitioning into machine learning from non-traditional careers.
In this episode, Richie and Marina explore how AI is reshaping the machine learning engineer role, the shifting balance between coding and planning, why evaluation matters more than ever, the differences between ML engineer, AI engineer, and data scientist roles, how to break into the field from a non-technical background, what makes a strong portfolio project, the hiring process at big tech, how to prepare for technical interviews, networking strategies that actually work, what success looks like in your first few months on the job, and much more.
Links Mentioned in the Show
• Chip Huyen — AI Engineering (book)
• Andrew Codesmith on YouTube
• Phillip Choi on YouTube
• A Life Engineered on YouTube
• Keras
• LeetCode
• Connect with Marina: LinkedIn
• AI-Native Course: Intro to AI for Work
• Related Episode: How to Have a Career in Data Science in 2025 with Dawn Choo
New to DataCamp?
Learn on the go using the DataCamp mobile app - https://www.datacamp.com/mobile
Empower your business with world-class data and AI skills with DataCamp for business - https://www.datacamp.com/business

Every conversation about AI in data eventually arrives at the same question: which roles survive, and which ones get automated away? Generative AI can already draft SQL, build dashboards, and run exploratory analysis — but it still can't sit with a business stakeholder and untangle what "customer" actually means across five teams. For data professionals, that shifts the day-to-day from production work toward translation, modeling, and judgment. So which skills are worth doubling down on? Which roles are becoming central, and which are quietly disappearing? And what should anyone hiring — or being hired — be paying attention to right now?
Veronika Durgin is the VP of Data at Saks Global, where she leads data strategy across the luxury retail group. A full-stack data executive with more than two decades of experience spanning database administration, data engineering, platform architecture, data modeling, and analytics, Veronika is a Snowflake Data Superhero and a member of CDO Magazine's Global Editorial Board. She writes about data modeling, data culture, and data leadership on her Substack and Medium.
In the episode, Richie and Veronika explore the future of data careers under AI, why analytics engineering becomes the catch-all role, the skills and hiring shifts data leaders are making, centralized data with decentralized analytics, keeping enterprise data teams agile, conceptual data modeling as the unglamorous prerequisite to AI, semantic layers, agentic commerce, and much more.
Links Mentioned in the Show:
Connect with Veronika: LinkedIn
Veronika's Substack: Think. Solve. Repeat.
dbt — referenced as the origin of "analytics engineering"
Open Data Science Conference (ODSC) — Veronika's recent talk on data and company politics
Amazon "two-way door" decisions — Bezos shareholder letter
Jessica Talisman — Veronika's recommendation for knowledge graphs and ontologies
Juan Sequeda — referenced on semantic layers and knowledge graphs
Catalog & Cocktails podcast (hosted by Juan Sequeda)
AI-Native Course: Intro to AI for Work
Related Episode: Creating an AI-First Data Team with Bilal Zia, Head of Data Science & Analytics at Duolingo

New to DataCamp?
Learn on the go using the DataCamp mobile app

Empower your business with world-class data and AI skills with DataCamp for business

Most AI ethics conversations sound the same: be fair, be transparent, be accountable. The values are right, but in practice they don't get teams out of bed in the morning. Executives nod along, employees take the compliance training, and meanwhile real risks like hallucinations, cascading failures, and autonomous agents acting at scale slip through. So what shifts when teams stop chasing an ethical ideal and start naming the specific disasters they want to avoid? Who needs to be in the room to spot them? And what kind of training actually changes how people use AI day to day?
Reid Blackman is the founder and CEO of Virtue, an AI ethical risk consultancy, and the author of The Ethical Nightmare Challenge: How to Avoid the Worst of AI (2026) and Ethical Machines (HBR Press, 2022). A former philosophy professor at Colgate with a PhD from the University of Texas at Austin, he has designed responsible AI programs for organizations including Amazon, Etsy, Kraft Heinz, Merck, US Bank, and Nationwide, and has advised the FBI, NASA, the World Economic Forum, and the Canadian government on federal AI regulations. He also hosts the Ethical Machines podcast.
In the episode, Richie and Reid explore why responsible AI fails to motivate organizations, the biggest AI ethical nightmares facing companies today, the unique risks of agentic AI including cascading failures and emergent risks, the Ethical Nightmare Challenge framework, cross-functional ENC teams, training employees in plain language, scaling AI governance, measuring success by what you avoid, and much more.
Links Mentioned in the Show:
• The Ethical Nightmare Challenge by Reid Blackman
• Ethical Machines by Reid Blackman
• Ethical Machines podcast
• Claude Code
• Connect with Reid: LinkedIn
• AI-Native Course: Intro to AI for Work
• Related Episode: #350 How to Make Hard Choices in AI with Atay Kozlovski
New to DataCamp?
Learn on the go using the DataCamp mobile app.
Empower your business with world-class data and AI skills with DataCamp for business.

Valerie Tiberius is the Paul W. Frenzel Chair in Liberal Arts and Professor of Philosophy at the University of Minnesota. She is an expert in ethics, moral psychology, and well-being, and the author of five books including What Do You Want Out of Life? and the forthcoming Artificially Yours: Real Friendship in a World of Chatbots (Princeton University Press, May 2026). She previously served as President of the Central Division of the American Philosophical Association.
In the episode, Richie and Valerie explore the purpose of friendship and whether AI can replicate it, the benefits and risks of chatbot companions for loneliness, how sycophantic AI responses distort advice and self-perception, the dangers of companion chatbots for children's social development, designing ethical AI companions that promote human flourishing, the zone of proximal development as a framework for better AI tools, and much more.
Links Mentioned in the Show:
Artificial Intimacy by Sherry Turkle
Being You: A New Science of Consciousness by Anil Seth
Liberation Day: Stories by George Saunders
Hard Fork podcast (NYT)
Connect with Valerie
AI-Native Course: Intro to AI for Work
Related Episode: #342 — "The Secrets to High AI Adoption" with Stefano Puntoni, Professor at Wharton

New to DataCamp?
Learn on the go using the DataCamp mobile app

Empower your business with world-class data and AI skills with DataCamp for business

Almost every AI agent demo lands in roughly the same place: it works most of the time, looks remarkable, and then fails in a way no one anticipated. Self-driving cars hit this wall a decade ago, and agents are running into it now. For data and AI teams, the question is no longer whether agents can complete a task — it's whether they can complete it reliably enough to remove the human reviewer. Which categories of work tolerate a 90% success rate? Which absolutely don't? And where should the next layer of guardrails sit?
Ruslan Salakhutdinov is a UPMC Professor of Computer Science at Carnegie Mellon University and one of Geoffrey Hinton's former PhD students. He has previously served as Director of AI Research at Apple and VP of Research in Generative AI at Meta. His research focuses on deep learning, reasoning, and AI agents.
In the episode, Richie and Russ explore the most exciting use cases of AI agents today, long horizon tasks, the credit assignment problem, multi-agent systems, designing reliable human-in-the-loop workflows, agent safety and guardrails, embodied and physical AI, lessons from self-driving cars, the difference between academia and industry, and much more.
Links Mentioned in the Show:
• Claude Code (Anthropic)
• Yutori
• Waymo
• Apple Project Titan
• DeepSeek-V3 Technical Report
• Kimi K2 Technical Report
• Connect with Ruslan: LinkedIn
• AI-Native Course: Intro to AI for Work
• Related Episode: AI Agents at Work: What Actually Breaks (and How to Fix It) with Danielle Crop
New to DataCamp?
Learn on the go using the DataCamp mobile app

Empower your business with world-class data and AI skills with DataCamp for business