Projects — Bhav Jain

Model safety / illustration

01 / LLM safety · adversarial testing · evaluation

Model safety research & red-teaming (selected)

Contract AI safety researcher and red-teamer. Passed technical screening and secured paid work testing frontier models; identified universal jailbreak-style vulnerabilities and helped drive mitigations pre-deployment.

Résumé (PDF)

NLP research / illustration

02 / NLP · prompting · evaluation

Town Hall Debate Prompting (publication)

Peer-reviewed NLP work on multi-persona interaction as a prompting method to improve logical reasoning in LLMs (vs. standard chain-of-thought baselines).

“Town Hall Debate Prompting: Enhancing Logical Reasoning in LLMs through Multi-Persona Interaction.”

Proceedings

Patent / illustration

03 / Materials engineering · experimentation · patent prosecution

Utility patent pending — novel bioplastic formulation

Independently designed and engineered a novel bioplastic formulation and filed a full utility patent with the USPTO; currently responding to office action.

View patent publication

Illustration · not investment returns

04 / Python · backtesting · statistics

Algorithmic trading research stack (supplementary)

Built and backtested systematic equity strategies in Python (momentum-driven signals, asymmetric risk construction), plus an end-to-end research workflow for iteration and evaluation.

All repositories on GitHub