Lukas Biewald on Tinkering, Observability, and Building Two AI Companies [AI Tinkerers - "One-Shot"] .

Lukas Biewald on Tinkering, Observability, and Building Two AI Companies

Alisa Delikatna
Alisa Delikatna — AI Tinkerers - "One-Shot"
December 18, 2025

(Banner) A promotional graphic, likely a video thumbnail, featuring a headshot of a man (Lukas Biewald) against a dark background with glowing circuit board patterns, advertising a technical interview. Text: LUKAS BIEWALD ON TINKERING, OBSERVABILITY, AND BUILDING TWO AI COMPANIES AI TINKERERS TECHNICAL INTERVIEW AI TINKERERS "ONE-SHOT" Bold, high-contrast, and technical/futuristic. | Colors: #FFFFFF, #000000, #1976D2, #708090 Note: The image is designed as a high-impact promotional graphic or thumbnail, combining large headline text and a photo subject against a stylized background, fitting the purpose and structure of a banner.
What happens when a lifelong tinkerer turns curiosity into two foundational AI companies?

In this episode of AI Tinkerers One Shot, we sit down with Lukas Biewald—founder of Weights & Biases and CrowdFlower (now Figure Eight)—to dissect how early projects forged the engineering mindset required to build essential tools for modern AI development.

We explore how the ethos of tinkering, starting with robot cars and Raspberry Pi experiments, translates directly into fostering a culture of experimentation inside fast-growing teams today.

The conversation dives deep into the realities of shipping AI in 2025, focusing on the practical challenges and breakthroughs builders face right now.

Watch the Episode

Watch now


What Builders Will Take Away

This episode pulls apart the choices and trade-offs made while scaling tools that underpin the AI industry. Lukas shares where AI tools are heading, what’s working in practice, and why curiosity still drives the best breakthroughs.

  • Agentic Coding: Opportunities and Challenges: A candid look at the state of agentic coding, including team structure and workflows that maximize agent output while maintaining code quality.
  • Observability in Production: Why observability is non-negotiable for AI systems in the real world, and the tooling required to prevent regressions and confidently scale complex LLM applications.
  • The Future of Fine-Tuning and RL: Where reinforcement learning and fine-tuning are headed, and what that means for builders looking to customize frontier models.
  • Lessons from Building Community-Driven Products: Practical insights from scaling both CrowdFlower (data labeling) and Weights & Biases (MLOps) by focusing on the needs of active practitioners.
  • Tinkering Culture: How creative play—including vibe coding and 3D printing with kids—opens doors for future engineers and keeps the experimental mindset sharp.

— 

💡 Resources

Lukas Biewald – https://www.linkedin.com/in/lbiewald/

AI Tinkerers – become a member: https://aitinkerers.org


Chapters

Time Topic
00:00 Intro & Guest Background
02:07 Early Tinkering and CrowdFlower
03:06 Building Robot Cars and Meeting Pete Warden
08:41 From Tinkering to Weights & Biases
12:27 Parenting, Vibe Coding, and Kids as Makers
21:56 3D Printing and Creative Play
24:25 AI Tools, Team Structure, and Company Growth
26:43 Agentic Coding: Opportunities and Challenges
35:40 AI in Production: Observability and Real-World Use Cases
49:28 The Future of AI, Fine-Tuning, and RL

Ready for more?

Check out other posts from this blog.

View all posts