My research interests are ML for materials science and chemistry, particularly generative models (DiffSyn), RL post-training for LLM scientific reasoning (SynReason), and materials synthesis (ZeoSyn). I received my PhD at MIT and Bachelors at Imperial. Beyond AI for Science, I worked as an AI Researcher at Meta and Google Research on diffusion models and transformers. I am honored to receive the MIT Presidential Fellowship for my PhD, my research featured on MIT News, and my scientific commentary highlighted on Yahoo finance.

	Massachusetts Institute of Technology PhD Researcher | Cambridge, MA Sep 2021 – Present Advisor: Elsa A. Olivetti Collaborators: Rafael Gomez-Bombarelli, Yuriy Roman-Leshkov, Manuel Moliner, Jennifer Rupp MIT Presidential Fellowship Young NUS Fellow
	Meta AI Research Scientist | Sunnyvale, CA June 2025 – Aug 2025 Foundational AI and Codesign team (internship): Self-feedback for transformers
	Google Research Research Scientist | Mountain View, CA + Cambridge, MA June 2024 – Jan 2025 SciML team (internship): Hierarchical controllable diffusion models
	Ro5 ML Engineer | London, UK Apr 2021 – Jul 2021 GNNs + RL for molecular generation
	Institute of High Performance Computing Research Engineer | Singapore Jul 2020 – Jul 2021 Classical ML for catalysis
	Imperial College London Research Assistant | London, UK Mar 2020 – Mar 2021 Constrained RL for process optimization

• [Featured on MIT News] Generative diffusion models for materials synthesis planning using molecular and crystalline materials datasets (Nature Computational Science, 2026 | NeurIPS (Oral Spotlight) AI for Materials, 2024 | Code) Elton Pan, Soonhyoung Kwon, Sulin Liu, Mingrou Xie, Alexander J Hoffman, Yifei Duan, Thorben Prein, Killian Sheriff, Yuriy Roman-Leshkov, Manuel Moliner, Rafael Gomez-Bombarelli, Elsa Olivetti

• RL post-training for synthesis reasoning (ICLR AI for Materials, 2026 | Code under progress) Elton Pan^*, Thorben Prein^*, Juno Nam, Xiaochen Du, Soojung Yang, Pengfei Cai, Jennifer L.M. Rupp, Rafael Gomez-Bombarelli, Elsa Olivetti

• RL-aligned flow matching for molecular crystal structure prediction (ArXiv, 2026 | Code under progress) Akshay Subramanian, Elton Pan, Juno Nam, Maurice Weiler, Shuhui Qu, Cheol Woo Park, Tommi S. Jaakkola, Elsa Olivetti, Rafael Gomez-Bombarelli

• LLMs for materials synthesis planning for inorganic materials (ACS Applied Materials & Interfaces | NeurIPS (Oral Spotlight) AI for Materials, 2025 | Code) Thorben Prein, Elton Pan, Janik Jehkul, Steffen Weinmann, Elsa A Olivetti, Jennifer LM Rupp

• LLM-enabled Bayesian optimization for molecular / protein optimization (NeurIPS (Spotlight) AI for Science, 2025 | Code) Mattias Akke, Soojung Yang, Jurgis Ruza, Jinyeop Song, Elton Pan, Rafael Gomez-Bombarelli

• Transformer-based ranker for synthesis precursor recommendation for inorganic materials (Paper | Code in progress) Thorben Prein, Elton Pan, Sami Haddouti, Marco Lorenz, Janik Jehkul, Tymoteusz Wilk, Cansu Moran, Menelaos Panagiotis Fotiadis, Artur P Toshev, Elsa Olivetti, Jennifer LM Rupp

• Model explainability/interpretability (Aggregated SHAP) for materials synthesis (ACS Central Science, 2024 | Code) Elton Pan, Soonhyoung Kwon, Zach Jensen, Mingrou Xie, Rafael Gómez-Bombarelli, Manuel Moliner, Yuriy Román-Leshkov, Elsa Olivetti

• Reinforcement learning (deep Q-learning, policy gradient) for inverse design of inorganic materials (NeurIPS AI for Materials, 2022 | npj Computational Materials, 2024 | Code) Elton Pan^*, Christopher Karpovich^*, Elsa Olivetti

• Reaction Graph Networks for modeling precursor-target interactions to predict materials synthesis routes (NeurIPS AI for Materials, 2024 | Code in progress) Thorben Prein, Fuzhan Rahmanian, Kesava Prasad Arul, Jasmin El-Wafi, Menelaos Panagiotis Fotiadis, Jan Heimann, Paul Weinmann, Yifei Duan, Elton Pan, Elsa Olivetti, Jennifer LM Rupp

• Synthesis condition prediction for inorganic materials (Chemistry of Materials, 2023 | Code) Christopher Karpovich, Elton Pan, Zach Jensen, Elsa Olivetti

• Materials representation learning (multi-task transformer pretraining) for inorganic materials property/synthesis prediction (NeurIPS AI for Materials, 2023 | Code in progress) Thorben Prein^*, Elton Pan^*, Tom Doerr, Elsa Olivetti, Jennifer Rupp

• Constrained RL for process optimization (Computers & Chemical Engineering, 2021 | Code) Elton Pan, Panagiotis Petsagkourakis, Max Mowbray, Dongda Zhang, Ehecatl Antonio del Rio-Chanona