Senior Knowledge Engineer, Ontology and Data Standards (Contract)

Altos Labs•San Francisco, CA

17d

About The Position

As a contract Senior Knowledge Engineer, Ontology and Data Standards, you will drive how our team represents biological knowledge. Altos has built the machinery: a large canonical LinkML/Biolink schema including REST/MCP/CLI service layer; specification packages covering multiple prior-knowledge sources; and an agentic / human-in-the-loop curation framework. This role requires a technical owner to govern and administer the content. You will serve as the scientific translation layer between biological discovery and AI platform engineering. Your responsibilities include eliciting implicit models from bench and computational scientists, encoding them rigorously, and utilizing agentic workflows that scale curation beyond manual review. This role carries explicit decision rights, including review authority on canonical schema change and ownership of the curation conventions that data atlas and engineering teams follow.

Requirements

PhD with 5+ years, or MS with 8+ years, in Bioinformatics, Computational Biology, Genomics, Systems Biology, or a related life science field.
Direct experience turning messy, real-world exploratory datasets (genomics, single-cell, screening, imaging) into structured metadata.
Strong working knowledge of OBO Foundry standards, OWL, RDF, and SPARQL, with schema modeling experience in LinkML, Biolink, or BioCypher.
Proficient in modular Python (3.12, uv, pandas, Pydantic, pytest), and able to read and extend a substantial existing codebase rather than only scripting against it.
Active hands-on experience with AI environments (e.g., Cursor, Claude Code) and building agentic LLM workflows, custom MCP tools, or RAG pipelines.
Familiarity with Git, GitHub Actions, graph/relational databases, and cloud storage (e.g., AWS S3).

Nice To Haves

Experience with Protégé, ROBOT, ODK, or SSSOM mapping sets.

Responsibilities

Design, maintain, and extend internal LinkML / Biolink models and ontologies covering experimental and prior-knowledge metadata—driven by competency questions from user requirements and not by source-side convenience.
Own mapping and alignment as a discipline: cross-ontology mappings with recorded justification and provenance, and upstream term requests as appropriate.
Serve as reviewer of record for schema change, holding the line that reuse precedes coining anything new.
Administer, steer, and refine our agentic curation framework and Model Context Protocol (MCP) service layer to scale data profiling, validation, and entity mapping far beyond manual capacity.
Standardize tool definitions (FastMCP), prompt templates, and LLM-driven workflows that extract, normalize, and validate metadata from prior-knowledge sources and experimental assets.
Define mapping patterns and metadata conventions for diverse experimental modalities (genomics, single-cell, screening data) to conform to the canonical schema.
Establish standardized conventions that allow both external prior-knowledge sources and internal experimental datasets to cleanly map into the LinkML graph architecture.
Establish and maintain automated validation, LLM-assisted quality control, and CI/CD under a "curation as code" paradigm—extending existing drift and coverage gates into the places where semantic review is still manual.
Measure whether the knowledge is correct, not merely reproducible: gold-standard sets, inter-curator agreement, and precision/recall on automated extraction.