Meet Thibault Sottiaux, Codex lead on the Right way to code, future of education, building sand box, security, coding languages to learn and much more! Thank you OpenAI for the opportunity!📲 Get premium wireless for as low as $15 a month with my partner at https://www.mintmobile.com/singhMy Recommendations: https://linktree.com/singhinusa1. How I Edit podcast & enhance audio: https://adobe.prf.hn/click/camref:1100l568ad/destination:https%3A%2F%2Fpodcast.adobe.com2. Checkout KodeKloud & Get upto 45% Off https://bit.ly/4dpsE7y (includes AWS, GCP, Azure Subscriptions with the Pro Plan)3. 📈 Checkout Moomoo from Palo Alto: Get 8.1% APY + 15 Free Stocks: https://start.moomoo.com/singhinusa (Legal of F1 VISA)4. 🏦 Adro US Account with No Incoming Wire Fees (Get $50) http://secure.joinadro.com/signup?referralCode=Singh5. 🎓SoFi Student Loan Refinance and you can get a $300 welcome bonus: https://www.sofi.com/invite/student-loans?gcp=174b310b-f306-454b-86e5-062ce06a9ec7&isAliasGcp=false0:00 – Introduction & CODEX Build • How Codex was built in 7 weeks, 630,000 PRs, async coding agents2:07 – Codex Model Improvements • New model, code quality, autonomy, embeddings, file relationships3:02 – How Programmers Use AI • Optimizing for understanding, junior vs senior devs, learning fundamentals4:10 – Key Metrics for AI-Driven Coding • Understanding codebase, prompting, planning features, learning tradeoffs6:21 – Curiosity & OpenAI Culture • Curiosity as a value, speed of development, bottom-up approach7:14 – Codex Engineering Sprint • Team structure, mergeability, code review, bottlenecks10:17 – AI Usage at OpenAI • Codex as collaborator, code review, safety net, public model usage11:18 – Impact of AI on Coding Culture • Bug finding, code review depth, contract vs behavior, timesaving13:06 – Parallel Agents & Collaboration • Using multiple Codex instances, context switching, future improvements14:11 – Codex in Developer Learning Journey • Accelerating language learning, idea realization, impact on career16:47 – Codex for Legacy Code & Refactoring • Working with old codebases, safe changes, porting languages18:13 – Automated Testing & Agents.md • Porting C++ to Rust, encoding expectations, agent preferences20:07 – Personalization & Steerability • Agents.md for code style, user preferences, avoiding cookie-cutter code21:03 – Optimizing for Learning with AI • Asking questions, curiosity, challenging understanding23:31 – Autocomplete vs Agentic AI • Codex as an agent, exploring environment, measuring success25:28 – Measuring Success & Impactful Code • Success criteria, better code, economic value, not just volume27:24 – Future of Coding Education • Bootcamps, AI as collaborator, evolving tools, critical thinking29:22 – Critical Thinking & Good Taste • Importance of taste, user understanding, iteration speed30:30 – Codex in Competitions & New Problems • Solving novel problems, gold medals, accelerating discovery32:05 – Security, Production Readiness • Secure, production-grade code, code review, test coverage33:15 – AI Engineer Skills & Tech Stack • Python, Rust, security, building applications, user understanding36:02 – Sandbox & Security Challenges • Agent safety, network policies, user approvals41:56 – Running Codex in Sandbox • Restricting access, safe defaults, supervision42:49 – AI Creating New Programming Paradigms • Potential for new languages, scientific discovery44:13 – Future Scientific Discovery • Health, material sciences, ecological advances45:36 – Closing Thoughts
