Simon Henniger

Computer Science PhD student at Harvard

✉ shenniger@g.harvard.edu

Hi, I am Simon.

I am building computational reasoning systems using a blend of LLMs and logic systems. Particular interests include:

• LLM-assisted Theorem Proving and Verified Programming (especially Secure Program Synthesis)
• Harness Engineering and Programming Languages for AI
• General AI Reasoning and its failure modes

I am a PhD student in Harvard's metareflection lab, and hold a Bachelor's degree in Computer Science from TU Munich, as well as Bachelor's and Master's degrees in Philosophy from the University of Munich. Previously, I interned at EPFL and Jane Street.

My studies are funded by the Harvard SEAS Prize Fellowship.

Please reach out – I like getting email.

Interests outside of Research

Below are some of my interests outside of my AI research, to serve as conversation starters if you like:

• Programming languages (including compiler construction)
• Entrepreneurship
• Philosophy (especially Vienna Circle and Nietzsche)
• Literature
• Game Design

Recent Preprints

• Gabriel Poesia, Simon Henniger, Tzu-Han Hsu, Yilun Du, and Nada Amin, “Formal Disco: Scalable Open-Ended Generation of Formally Verified Programs”, July 2026.
• Simon Henniger* and Gabriel Poesia*, “The Token Games: Evaluating Language Model Reasoning with Puzzle Duels”, February 2026.
• Roy Rinberg, Annabelle Michael Carrell, Simon Henniger, Nicholas Carlini, and Keri Warr, “Haiku to Opus in Just 10 bits: LLMs Unlock Large Compression Gains”, February 2026.
• Arnau Marin-Llobet, Simon Henniger, and Mahzarin R. Banaji, “Vision-Language Models Suppress Female Representations Under Ambiguous Input”, May 2026.

Publications

• David Brandfonbrener, Simon Henniger, Sibi Raja, Tarun Prasad, Chloe Loughridge, Federico Cassano, Sabrina Ruixin Hu, Jianang Yang, William E. Byrd, Robert Zinkov, and Nada Amin, “VerMCTS: Synthesizing Multi-Step Programs using a Verifier, a Large Language Model, and Tree Search”, in MATH-AI, the 4th Workshop on Mathematical Reasoning and AI, held at NeurIPS 2024. Outstanding Paper Award.
• Simon Henniger and Nada Amin, “The Dolorem Pattern: Growing a Language through Compile-Time Function Execution”, in 37th European Conference on Object-Oriented Programming (ECOOP 2023).