Staff Engineer, Chemical Search

About the role

This first cheminformatics hire will design and build the chemistry search engine: the system that exposes our accessible chemical space through query modes fit for how our partners actually design molecules and run their discovery campaigns. This engine will be searched by generative models, AI agents, and medicinal chemists alike. The engine is grounded in predictive models that capture the actual capabilities of our platform rather than a static catalog, which changes the underlying architecture in ways an experienced builder in this space will already have opinions about. You will take in constraints from our automation team, conversion and substrate-scope models from our ML team, and building blocks from our vendors to define the complete specification of Satomic's accessible chemical space. You will also translate insights from our partners' discovery strategies into fit-for-purpose query modes, keeping the engine in sync as both Satomic's accessible chemistry expands and our partners' discovery workflows evolve.

Qualifications

• Track record of owning a production chemistry search (or closely analogous retrieval / indexing system over structured scientific data) end-to-end in production, with real users on a customer-facing critical path. You've delivered sub-second p95 for interactive query workloads and sustained high-throughput batch scoring (think 10k+/second), and you know which architectural calls get you there. You think in hot paths, indexing strategies, and inference at scale, and make decisions that survive both growth and change in the underlying domain.
• Production-level fluency with core cheminformatics tooling (RDKit and/or OpenEye), molecule and reaction representations, with strong opinions about when each is the right primitive.
• Working experience with at least one major retrosynthesis system in production: you have lived with their pains and shortcomings, know where they break, and can articulate what a retrosynthesis layer for a defined-capability platform should do differently, specifically when narrowly scoped against validated reactions.
• Direct experience with generative chemistry methods, or the confidence and track record to become an expert on generative chemistry sampling and how accessible-space engines are best exposed to it.
• Enough chemistry intuition to hold a technical conversation with a medicinal or synthetic organic chemist as a peer.
• Confidence making build-vs-buy decisions across third-party cheminformatics tooling, and a track record of integrating, extending, or deliberately replacing it.
• Solid engineering fundamentals with room to grow: Python proficiency required, paired with a genuine enthusiasm for software engineering. We're looking for someone curious about expanding their toolkit and is excited to pick up new languages and explore where different tools shine, whether that's a performance-oriented language like Rust, Go, or C++ or something else entirely. Familiarity with data stores or vector indexes (FAISS, ScaNN, pgvector, or similar) is welcome, though you'll have opportunities to deepen this skillset.