
Diogo Almeida is the Co-founder and CEO of TypeSafe AI, a company building a new class of large language models for true no-human-in-the-loop automation. Diogo brings a decade of deep AI research experience and was formerly at OpenAI and Google Brain. He played a foundational role in shaping the modern LLM landscape, credited with co-inventing InstructGPT, RLHF, ChatGPT, and GPT-4. His work at TypeSafe AI focuses on rethinking foundational LLM assumptions to prioritize reliability, latency, and cost, to build AI that can be embedded deep inside real systems.
Diogo Almeida co-invented InstructGPT and RLHF at OpenAI (the methods behind ChatGPT) and co-authored the GPT-4 Technical Report, and is now asking what the next generation of AI must look like to run real systems, not just conversations. As co-founder and CEO of TypeSafe AI, his session offers a rare insider perspective on why the gap between AI's promise and real-world automation persists, and how TypeSafe is rebuilding the AI stack from first principles for machine-to-machine reliability.
Public activity researched automatically · as of Jun 2026