Remote AI Research Scientist: LLMs for Healthcare

GenMD; United States; Remote; USD 6,300 - 11,000; Full time

Benefits offered by this job

Fully remote

CA hours overlap

GPU servers access

Stanford collaboration

Publish research papers

Job summary

GenMD, a Stanford spin-off focused on AI and data infrastructure for healthcare, seeks an AI Research Scientist to work on LLMs, synthetic data generation, model evaluation, and scalable training. Collaboration with Stanford researchers and access to extensive clinical data are key highlights.

You will train and fine-tune LLMs, build evaluation pipelines, and explore MoE and ensemble techniques, with strong emphasis on reproducibility and real-world healthcare impact.

Qualifications

• Strong hands-on expertise with LLMs.
• B.Sc. (minimum) in CS, AI, ML, NLP or related field, thesis work in LLM/NLP preferred.
• Publications in LLM/NLP are strongly preferred.
• Strong Python skills with PyTorch, Hugging Face, model training, fine-tuning, and evaluation.

Responsibilities

• Train, fine-tune, and optimize LLMs using PyTorch and Hugging Face.
• Build synthetic data generation pipelines at scale.
• Develop LLM benchmarks, automated evaluation pipelines, and LLM-as-a-Judge systems.
• Experiment with supervised, semi-supervised, and unsupervised learning.
• Explore Mixture-of-Experts, model ensembles, distillation, and other modern LLM techniques.
• Translate research ideas into reproducible experiments and production systems.

Skills

LLMs

Python

PyTorch

Hugging Face

Model training

Fine-tuning

Evaluation

NLP

Education

B.Sc. in CS/AI/ML/NLP

Tools

Python tooling

Job description

GenMD, a Stanford spin-off focused on AI and data infrastructure for healthcare, seeks an AI Research Scientist to work on LLMs, synthetic data generation, model evaluation, and scalable training. Collaboration with Stanford researchers and access to extensive clinical data are key highlights.

You will train and fine-tune LLMs, build evaluation pipelines, and explore MoE and ensemble techniques, with strong emphasis on reproducibility and real-world healthcare impact.