Boris Cherny is the key person behind Claude Code at Anthropic, drawing on experience and insights gained through a rapid rise to Principal Engineer at Meta (Facebook, Instagram). This video vividly covers the diverse projects and promotion episodes he experienced at Meta, the lessons he learned resolving conflicts between organizations, and a deep discussion about how engineers should work in the 2025 AI era and the productivity revolution Claude Code has brought.

1. Starting at Meta: Growth as a Generalist and Latent Demand

Boris recalls the early days of his career at Meta, mentioning the "Chats in Groups" project connecting Facebook Groups with Messenger. He wasn't just coding -- he was a generalist handling everything from user research (UXR) to design. When the team had no researcher, he'd go to the company cafeteria and have employees test features himself.

Engineering seems like a very narrow skill set, but what you're actually doing is building products or infrastructure. There's so much more beyond writing code needed to carry out the entire process. I really love working with these generalists.

He identified "latent demand" as the most important principle in product development -- observing how users "hack" products in unintended ways to discover new product opportunities. Facebook Marketplace and the dating app both started from these kinds of observations.

I think latent demand is the single most important principle in product. (...) As a product principle, you can't force people to do behavior they're not already doing. What you can do is discover the intent they already have and channel it better.

2. Cultural Differences Between Organizations and Collaboration Challenges

Boris was candid about the extreme cultural differences he experienced working between the Facebook app team and the Messenger team. The Facebook team's philosophy was "ship products fast," while the Messenger team prioritized "stability and performance." This led to mutual distrust and conflicting goals -- a "nightmare" scenario.

Saying "difficult" is too restrained. It was really a nightmare. (...) For companies or organizations with fundamentally different values to collaborate successfully, you have to find goals or hypotheses that are interesting and shareable for both organizations.

He reflects that resolving these structural issues required either aligning incentives between organizations or changing the organizational structure itself (e.g., moving the Messenger team into the Facebook organization).

3. Side Quests and Engineering Influence

Boris credits "side quests" -- side projects beyond his main work -- as one of the secrets to his rapid growth. He constantly explored new things with curiosity, creating the React state management library "Undux" and writing a TypeScript book.

He particularly grew his influence through "better engineering" -- automating problems he encountered repeatedly (creating lint rules, etc.) and spreading these to other engineers, exercising technical leadership as the organization grew from 150 to 800 people.

Better engineering is the easiest way to expand your network and gain influence as an engineer. (...) When I deployed automation tools or lint rules to my teammates, it became leverage not just for me but for the entire team.

4. The Art of Promotion: From IC6 (Staff) to IC7 (Senior Staff)

Boris rose rapidly through promotions at Meta, with important lessons at each stage.

• IC6 (Staff) promotion: When the Facebook desktop redesign project "Comet" launched, he volunteered the Groups product early on, solving technical challenges first and building trust with the infrastructure team.
• IC7 (Senior Staff) promotion: He led critical data modeling decisions in the "Public Groups" project. Despite a technical disagreement with a senior engineer (Bob), he initially "disagreed and committed" to build trust, then later corrected the wrong decision based on data -- earning even greater trust and his promotion.

The most important thing is that you need to earn trust. (...) Until you earn trust, it's hard to demonstrate good technical judgment. Invest time building trust first.

He also shares an impressive anecdote about scoping a massive project with hundreds of engineers, where he organized a "technical design competition" -- splitting engineers into two teams to competitively find solutions over three hours, enabling rapid consensus on an uncertain technical direction.

5. Moving to Instagram and the 'Unshipping' Philosophy

After getting married and moving to Japan, Boris transferred to the Instagram team. There he learned Instagram's distinctive "craftsmanship" culture and the practice of "unshipping" features.

Unshipping means boldly removing features that don't meet certain usage thresholds. (...) Screen real estate is a limited resource, and every feature competes for it. (...) A small number of users might be upset, but on average, you're providing a much better experience for the majority.

With fewer meetings thanks to remote work, he was able to deeply immerse himself in coding again, and he proved his value once more by leading a massive technical debt resolution project migrating Instagram's old Python code to the modern Hack language.

6. Joining Anthropic and the Secret to Claude Code's Success

Boris joined Anthropic driven by his interest in science fiction and AI's future, as well as a sense of mission around AI safety. He was captivated by Anthropic's genuine commitment to building safe AI for humanity, not just developing technology.

The secret behind Claude Code's explosive internal reception was "building products for the model six months from now, not today's model."

Ben pushed me to think bigger. (...) "Don't build for today's model, build for the model six months from now." (...) So we predicted future performance and designed the product accordingly, and when the Opus 4 model came out, the product worked perfectly.

Currently, per-engineer productivity at Anthropic has increased by nearly 70% thanks to Claude Code. Beyond engineers, data scientists and even sales teams are also writing code and automating their work using Claude Code.

7. Coding and Productivity in the AI Era

Boris says that the way coding works has completely changed. He no longer types all the code himself -- instead he works by running multiple AI agents (Claude Code) simultaneously, orchestrating tasks.

Productivity tips used to be about how to allocate your time. But now it's different. "Learn how to use Claude Code, and learn how to run multiple Claudes simultaneously." That's the future of engineering.

Finally, he emphasizes focusing on solving user problems rather than worrying about competitors (Cursor, Copilot, etc.), and advises newcomers to follow "common sense." Don't get swayed by big-company bureaucracy or misaligned incentives -- use common sense to judge what users and the market want, and execute.

Conclusion

Boris Cherny's story goes beyond a simple success narrative, illustrating the attitudes engineers need to grow amid technological change. Product sense for identifying latent demand, side projects for proactively solving problems, leadership built on trust, and productivity innovation through AI tools -- this interview resonates deeply with anyone thinking about the path forward as an engineer in the rapidly evolving AI era of 2025.