Eric Zelikman

I'm CEO and co-founder of humans&. We believe we can build models to understand and empower people, instead of just to automate tasks. If you're passionate about this mission, reach out!

At xAI, I was an early contributor to pretraining data for Grok 2, kicking off and scaling RL for reasoning for Grok 3 Thinking, building the agent RL infra and recipe for Grok 4, then developing its first multi-agent RL recipes.

Before, I was a Stanford Ph.D. candidate, advised by Nick Haber and Noah Goodman. There, I wrote STaR, the first work teaching language models to reason in language using their own rationales.

Industry

  • Co-founder and CEO humans& (Sep 2025 onwards)
  • Member of Technical Staff xAI (Mar 2024 - Sep 2025)
  • Student Researcher MSR (Jun – Sep 2023); Blueshift @ X & Google Research (Jun 2022 – Sep 2022)
  • Deep Learning Engineer Lazard (Jul 2020 – Sep 2021)
  • Curriculum Developer DeepLearning.AI GANs (Jun – Oct 2020)
  • Machine Learning Intern Argo AI (Jun – Sep 2019); Uncountable (Jun – Sep 2018 & Apr – Jun 2019)

Education

Stanford University BS, Symbolic Systems w/ Honors (2020); former CS PhD student advised by N Haber and ND Goodman.

Academia

Selected Works

  1. Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking E Zelikman, G Harik, Y Shao, V Jayasiri, N Haber, N Goodman COLM 2024
  2. Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation E Zelikman, E Lorch, L Mackey, A Kalai COLM 2024 [Oral Spotlight], NeurIPS OPT Workshop 2023
  3. Hypothesis Search: Inductive Reasoning with Language Models R Wang*, E Zelikman*, G Poesia, Y Pu, N Haber, N Goodman ICLR 2024
  4. Self-Supervised Alignment with Mutual Information: Learning to Follow Principles without Preference Labels J Fränken, E Zelikman, R Rafailov, K Gandhi, T Gerstenberg, N Goodman NeurIPS 2024
  5. Certified Reasoning with Language Models G Poesia, K Gandhi*, E Zelikman*, N Goodman TMLR 2024
  6. Parsel 🐍: Algorithmic Reasoning with Language Models by Composing Decompositions E Zelikman, Q Huang, G Poesia, N Goodman, N Haber NeurIPS 2023 [Spotlight]
  7. Generating and Evaluating Tests for K-12 Students with Language Model Simulations E Zelikman*, W Ma*, J Tran, D Yang, J Yeatman, N Haber EMNLP 2023
  8. Lexinvariant Language Models Q Huang, E Zelikman, S Chen, Y Wu, G Valiant, P Liang NeurIPS 2023 [Spotlight]
  9. Just One Byte (per gradient): A Note on Low-Bandwidth Decentralized Language Model Finetuning Using Shared Randomness E Zelikman, Q Huang, P Liang, N Haber, N Goodman 2023
  10. STaR: Bootstrapping Reasoning With Reasoning E Zelikman*, Y Wu*, J Mu, N Goodman NeurIPS 2022
  11. Holistic evaluation of language models (reasoning) P Liang, R Bommasani, T Lee, D Tsipras, D Soylu, M Yasunaga, Y Zhang, D Narayanan, Y Wu, ... , E Zelikman, ... TMLR 2022
  12. Evaluating the Disentanglement of Deep Generative Models through Manifold Topology S Zhou, E Zelikman, F Lu, A Ng, G Carlsson, S Ermon ICLR 2021
  13. Learning is its Own Reward: Exploring Worlds with Curiosity-driven Spiking Neural Networks E Zelikman (advised by N Haber) Undergraduate Honors Thesis 2020
  14. Contextual Salience for Fast and Accurate Sentence Vectors E Zelikman, R Socher arXiv 2018