About the role
We are looking for a Forward Deployed ML Engineer with 2+ years of experience to sit at the intersection of cutting-edge AI and real-world oncology workflows at Triomics. This is not a research role — we need someone who has built and shipped multi-step LLM/agent pipelines in production, can own a customer delivery end-to-end, and is comfortable jumping on a call with a nurse, physician, or data science team to close accuracy gaps. You'll be the critical bridge between our AI research team and our clinical customers, translating complex clinical feedback into rapid pipeline iterations. If you've built agentic systems that actually run at scale and you're excited about applying that to cancer care, this role was made for you.
What will you be doing?
- Design and build agentic extraction pipelines that process 500+ page patient charts (clinical notes, pathology reports, imaging reports, genomic panels) and output structured JSON per customer data dictionaries
- Own accuracy end-to-end: define evaluation datasets, run precision/recall analysis per variable, identify failure modes, and iterate through agent architecture changes, prompt engineering, fine-tuning, or rule-based post-processing
- Act as the technical bridge between customers and the internal AI research team — join customer calls, capture clinical feedback, and either make quick pipeline iterations yourself or translate requirements for the core ML team
- Go deep into real patient charts to understand how oncologists document data, why certain variables are ambiguous, and use that understanding to improve extraction accuracy
- Deliver on customer timelines through intense sprint periods, then iterate and improve — you manage your own workstreams across multiple customers simultaneously
