Nathan is a machine learning researcher who works on building, understanding, and advocating for open language models and other responsible autonomous systems.
He is currently a post-training lead at the Allen Institute for AI.
Papers → Google Scholar, my CV, Semantic Scholar, see PhD Thesis.
Talks → YouTube channel or CV.
Newsletter → Interconnects.ai
Book → RLHF Book
Podcast appearances
I host two podcasts (sporadically).
- Interconnects Interviews cover the latest technical trends in AI.
- The Retort covers the political economy and stakes of AI.
Otherwise, I sometimes appear on friends' podcasts:
- May 2025: ChinaTalk - Sycophancy, US-China Competition, RL agents, and some career advice.
- Feb. 2025: Lex Fridman - DeepSeek and what it means for AI.
- Jan. 2025: AI Summer - (a new podcast from friends Tim and Dean) discussing post-training and reasoning models.
- Dev. 2024: ChinaTalk - the biggest AI stories of the year and what it's like to train SOTA AI models.
- Nov. 2024:The Cognitive Revolution - a tour of the state-of-the-art in LM post-training (also on ThursdAI for that week).
- Jun. 2024: SuperDataScience - a general show covering trends in RLHF and robotics + LLMs.
- May. 2024: ChinaTalk - recent trends in AI policy, and of course China.
- Feb. 2024: Practically Intelligent - Discussing how to maintain a media diet in ML and cut through the hype.
- Jan. 2024: Latent Space - RLHF lecture and Q&A.
- Sep. 2023: Generating Conversation - RLHF and LLM evaluation.
- Dec. 2023: Sharon's AI Extra - predictions for 2024.
- Jul. 2023: Latent Space - Llama 2 launch.
- Jul. 2023: Practically Intelligent - evaluation and open-source LLMs.
- Mar. 2021: TalkRL podcast - model-based RL.
- Jan. 2021: Software Engineering Daily - mixed research roundup.
Guides
Some useful links for graduate students and academics.
- How to Review a Paper - My process for writing thorough paper reviews.
- Applying to Graduate School - Advice for prospective graduate students.
Popular Posts
Some of my most popular posts before transitioning to Substack (Top Posts there).
- Job Search as a PhD in ML - My experience on the AI job market.
- AI Research Job Market (Part 2) - Follow-up thoughts on the job market.
- Debugging Model-Based RL Systems - Practical debugging strategies.
Other
- Office Hours - I (sometimes) hold regular office hours through the ML Collective.
- Contact - Get in touch with me.
- Bookshelf - Books I've read and recommend.
- Buy me a ☕️ - Support my work.
- RLHF Book - My upcoming book on reinforcement learning from human feedback.