Clearly a strong senior AI/platform architect with excellent autonomous ownership, security/compliance expertise, and hands-on agentic AI work. The gap is that this role specifically requires deep ontology/knowledge-model and graph database experience. While knowledge graphs are mentioned, there is…
Outstanding infra and distributed systems engineer, but the experience aligns much more with platform engineering at large-scale infrastructure companies than with the semantic data platform focus we have.
Cortex is hiring a Chief Platform Engineer to own Guild — the entire data layer beneath its reasoning platform. This is a C-suite seat reporting to the CTO. You'd inherit and evolve the ontology, knowledge graph, and pipelines, set technical direction for the platform layer, and eventually hire and lead the platform team. On a three-person team raising again in August, the founders describe themselves as spread too thin — this hire takes the whole platform off the CTO's plate.
You'll design and maintain the ontology in TypeDB (TypeQL) — the entity types, relations, and rules that model experiments, assays, materials, and evidence — and own the knowledge graph and Qdrant vector store in production, including the hybrid symbolic-and-semantic retrieval the reasoning agents depend on. Agents query the graph but never mutate it; you define and enforce the audited write path. You'll build ingestion and entity-resolution pipelines that normalize messy, heterogeneous inputs (instruments, ELN/LIMS, publications, assay results) with provenance captured at write time, own platform security end to end, and lead SOC 2 — prepping for the GxP and 21 CFR Part 11 requirements that pharma partners bring.
What You'll Own
- The ontology in TypeDB/TypeQL — entity types, relations, and rules for experiments, assays, materials, evidence
- Knowledge graph + Qdrant vector store in production: schema evolution, query performance, hybrid symbolic + semantic retrieval
- Ingestion + entity-resolution pipelines normalizing heterogeneous inputs, with provenance at write time
- The controlled, audited write path to the symbolic layer — agents read, never mutate
- Platform security: authN/authZ (RBAC/ABAC), tenant isolation, secrets management, encryption in transit and at rest, audit logging
- SOC 2 certification; readiness for GxP / 21 CFR Part 11 data-integrity and audit-trail requirements
- SLOs and observability across the platform's FastAPI services
- Engineering standards, and hiring the platform team
- Ramp: onboard on internal docs in week 1 and get fluent fast on the data-interaction and multi-agent-harness sides; by end of month 1, take over development and evolution of the platform from the CTO
Requirements
- Designed and operated production ontologies or knowledge models — Palantir Foundry/Gotham, or a comparable semantic-layer, master-data, or knowledge-graph system
- Strong command of graph databases (TypeDB/TypeQL, Neo4j, or RDF/SPARQL) and data modeling
- Production Python backend; FastAPI or equivalent
- Data-integration pipelines built against messy, heterogeneous sources
- Applied security: authorization models, multi-tenant isolation, secrets, encryption, threat modeling — ideally on sensitive or regulated data
- Genuinely strong across both front end and back end
Execution and Ownership
- Sets architecture and operates autonomously at an early-stage company — this is the platform owner
- Obsession with product and real leadership instinct (they'll build and lead the team)
- Independent thinker who reasons from first principles, not just wiring together AI-assisted boilerplate
Background
- Deliberately wide band: either ~1–2 years out and hungry enough to grind and be molded, or a Palantir-level operator worth top dollar. CS plus formal education reads as a strong signal; where you last worked matters more than which school
- End-to-end personal projects, original CS research, or open-source work weighted heavily
Location and Visa
- In person in San Francisco
- Sponsorship open — the team has H-1B/L-1 counsel on call and believes the best talent is international
Nice-to-Have
- Vector search / retrieval (Qdrant or equivalent)
- Life sciences, biotech, or scientific-computing background
- Direct experience taking a system through SOC 2 or HIPAA
- Agent orchestration or probabilistic programming (DSPy, Pyro)
- Philosophy or formal-logic bent — the founders think in first- and second-order logic and are building a "reading room filled with philosophy"
- Hackathon wins, coding-comp medals, or athletics — reads as dynamic, not required