# Jack Gallifant

> Applied research lead at R37, R1’s AI lab. Agents and post-training for healthcare. Former AI researcher at MIT and Harvard Medical School and NHS physician.

Public profile from [Jack Gallifant’s website](https://www.jackgallifant.com). The biography below is written in the first person.

I believe AI can expand what medicine understands and what healthcare can deliver. That means advancing the science of care and removing the work that gets in its way.

I am an applied research lead at [R37](https://www.r1rcm.com/r37), R1’s AI lab.

My team and I build training and evaluation environments for long-horizon healthcare tasks, from clinical coding and [insurance appeals](https://www.r1rcm.com/resources/long-context-medical-reasoning-to-reduce-denials-post-training-nvidia-nemotron-3-5-super) to computer use and voice workflows.

We use these environments to post-train models. Our agents run in production at R1.

Previously, I was an AI researcher at MIT and Harvard Medical School. Before that, I worked as a physician in the NHS in London, UK.

My work has appeared in Nature Medicine, Nature Cancer and at NeurIPS. It has been covered by [The New York Times](https://www.nytimes.com/2025/10/30/well/chatgpt-health-questions.html) and STAT.

I’ve spoken at [TEDxBoston](https://www.youtube.com/watch?v=seUxBboIxDk), lectured at Harvard and MIT, and mentored teams at health AI datathons in 10 countries.

## Education

- MBBS, University of East Anglia
- MSc in physiology, King’s College London

## Papers and writing

- [Long-context medical reasoning to reduce denials: post-training NVIDIA Nemotron 3.5 Super VL](https://www.r1rcm.com/resources/long-context-medical-reasoning-to-reduce-denials-post-training-nvidia-nemotron-3-5-super): R1 blog, 2026. Related coverage: [NVIDIA](https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/).
- [An agentic AI system enhances clinical detection of immune-related adverse events in a multi-phase validation study](https://www.nature.com/articles/s43018-026-01264-6): Nature Cancer, 2026.
- [When helpfulness backfires: LLMs and the risk of false medical information due to sycophantic behavior](https://www.nature.com/articles/s41746-025-02008-z): npj Digital Medicine, 2025.
- [A Field Guide to Deploying AI Agents in Clinical Practice](https://arxiv.org/abs/2509.26153): arXiv, 2025.
- [The TRIPOD-LLM reporting guideline for studies using large language models](https://www.nature.com/articles/s41591-024-03425-5): Nature Medicine, 2025.
- [MedBrowseComp: Benchmarking Medical Deep Research and Computer Use](https://arxiv.org/abs/2505.14963): arXiv, 2025.
- [Sparse autoencoder features for classifications and transferability](https://arxiv.org/abs/2502.11367): EMNLP, 2025.
- [Cross-Care: Assessing the Healthcare Implications of Pre-training Data on Language Model Bias](https://arxiv.org/abs/2405.05506): NeurIPS, 2024.
- [Language Models are Surprisingly Fragile to Drug Names in Biomedical Benchmarks](https://arxiv.org/abs/2406.12066): Findings of EMNLP, 2024.
- [A closer look at AUROC and AUPRC under class imbalance](https://arxiv.org/abs/2401.06091): NeurIPS, 2024.

## Contact and profiles

If you’re advancing what AI can do for medicine—or putting it to work to improve care—I’d be glad to hear from you.

- [Google Scholar](https://scholar.google.com/citations?user=SlLz8KoAAAAJ)
- [LinkedIn](https://www.linkedin.com/in/jackgallifant/)
- [X](https://x.com/JackGallifant)
- [GitHub](https://github.com/Gallifantjack)

- [Email contact](https://www.jackgallifant.com/#contact): activate “Email me” on the website to reveal the address.