Truveta is looking to engineer the next wave of healthcare AI by developing adaptive agentic systems and fine-tuned foundation models that continuously learn, reason, and accelerate real-world impact, transforming how health data becomes intelligence to power breakthroughs for clinicians and researchers.
Requirements
- Proficiency with agentic AI frameworks (e.g., LangGraph, AutoGen, CrewAI) and understanding of interoperability protocols such as MCP and A2A.
- Hands-on experience fine-tuning and optimizing large language models (LLMs) or multimodal models, using techniques such as LoRA, PEFT, and TRL.
- Experience working with vector databases and embeddings, integrating FAISS, Pinecone, Chroma, or Azure AI Search into retrieval-augmented generation (RAG) or semantic-search pipelines.
- Strong software engineering fundamentals, with proficiency in Python and experience designing scalable systems in modern cloud environments (Azure, AWS, or GCP).
- Knowledgeable in transformer architectures, attention mechanisms, and tokenization principles.
- Understand how embeddings, context windows, and model scaling laws influence quality, cost, and performance.
- Fluent in ML validation and measurement (Precision, Recall, Specificity, NPV, etc.)
Responsibilities
- Design and reason with agentic AI frameworks — experienced in building multi-agent workflows using frameworks such as LangGraph, AutoGen, or CrewAI, integrating reasoning, planning, and memory to create intelligent, goal-driven systems.
- Bring a deep understanding of LLM fundamentals — knowledgeable in transformer architectures, attention mechanisms, and tokenization principles.
- Work fluently with embeddings and vector stores — experienced in building or integrating retrieval-augmented generation (RAG) pipelines, managing vector databases (e.g., FAISS, Pinecone, Chroma, Azure AI Search, or similar), and leveraging semantic search or context injection to enhance reasoning.
- Excel at model fine-tuning — with hands-on expertise fine-tuning large language models (LLMs) or multimodal models using supervised, reinforcement, or instruction-tuning techniques (e.g., LoRA, PEFT, TRL).
- Think and build like engineers — grounded in strong software design principles, modular architecture, and code quality.
- Collaborate across boundaries — partnering with platform, application, and product engineers to transform concepts into scalable, reliable AI solutions.
- Understand evaluation deeply — fluent in ML validation and measurement (Precision, Recall, Specificity, NPV, etc.), and experienced designing reward functions, evaluators, or grader models for reinforcement fine-tuning and continuous improvement.
Other
- 5+ years of experience building and deploying scalable, production-ready ML systems in a collaborative engineering environment.
- Demonstrate senior-level ownership — capable of setting technical direction, mentoring peers, and making pragmatic design trade-offs that balance innovation, performance, and reliability.
- Adapt and learn continuously — staying ahead of evolving AI architectures, agentic frameworks, and emerging paradigms in reasoning and retrieval.
- Act with purpose — applying thoughtful engineering and ethical AI principles to create systems that advance healthcare intelligence responsibly and at scale.
- All applicants must be authorized to work in the United States for any employer as we are unable to sponsor work visas or permits (e.g. F-1 OPT, H1-B) at this time.
