Research Engineer — Post-Training & Small Language Models (SLMs), Healthcare AI

$110,700 – $379,200 · Posted today

San DiegoLast seen today

Summary

Design and build specialized small language models (SLMs) for healthcare, with expertise in post-training techniques (SFT, RLHF, DPO, PPO, GRPO) and reasoning-model training. Develop healthcare-grade evaluation frameworks covering hallucination detection, factuality, bias/fairness, and clinical benchmarking, while performing red teaming and adversarial testing to identify alignment failures and unsafe behaviors. Apply PHI/HIPAA-aware data handling and produce model documentation for regulated clinical deployment. Work with open-weight foundation models (Llama, Qwen, Mistral, DeepSeek) using PyTorch and distributed training frameworks.

Data Science AI Machine Learning Python Adversarial Testing Agentic Systems Attention Mechanisms Bias Fairness Evaluation Data Filtering Decoding Strategies Deduplication Deepseek Deepspeed Dpo Evaluation Frameworks Factuality Evaluation Fsdp Grpo Hallucination Detection Hipaa Llama LLM Benchmarking Lora Megatron Lm Mistral Model Scaling Ollama Peft Phi Ppo Preference Optimization PyTorch Qlora Quantization Aware Training Qwen Ray Reasoning Models Red Teaming Reward Modeling Rl Rlaif Rlhf Rlvr Sft Synthetic Datasets Tensorrt LLM Tgi Tokenization Tool Use Transformer Architectures Vllm