Lilian Weng
Lilian Weng is an AI researcher whose work spans robotics, applied research, safety systems, and technical writing. At OpenAI, she contributed to the team that trained a robot hand to solve a Rubik’s Cube. In a 2022 interview, she described her later applied-AI work, and by July 2024 she was head of safety systems. She was among Thinking Machines Lab’s founding team in 2025. Her Lil’Log notes, begun in 2017, explain topics such as attention, diffusion models, and language-model agents with references to the underlying research. In July 2026, she wrote about harnesses for AI self-improvement.
- Now at
- Thinking Machines Lab
- Works
- 35
- Citations
- 3,791
- h-index
- 14
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
2017–2018
Public learning notes
Weng writes explanations of generative models and attention on Lil’Log.
-
2019
Robot manipulation
She coauthors the OpenAI robot-hand Rubik’s Cube paper with its research team.
-
2021–2023
Diffusion and agents
Her surveys explain diffusion models and planning, memory, and tools around language models.
-
2022–2024
Applied research and safety
She worked in applied AI and led safety systems.
-
2025
Thinking Machines launch
She was among the founding team.
-
July 2026
Research harnesses
Her latest technical post examines the systems around AI-assisted research and self-improvement.
Sources and credits
Photo: Girl Geek X event speaker profile. License: Not stated by source. Source: girlgeek.io. Cropped to a square and resized. Ask us to remove or replace this photo.
- Generative adversarial networks overview
- Attention? Attention!
- Solving Rubik’s Cube with a Robot Hand
- What are Diffusion Models?
- The power of continuous learning
- LLM Powered Autonomous Agents
- OpenAI Forum: AI Trainers Community Mixer
- Thinking Machines Lab: historical indexed launch record
- Harness Engineering for Self-Improvement