We are a small Sofia studio shipping agentic AI for production — evals, traces, durable execution, and the orchestration that keeps swarms honest past day 30.
Open roles are listed below. We hire senior engineers who like ownership, clear contracts, and boring reliability next to interesting models.
Remote-friendly within EU time zones; Sofia office available. We reply to every serious application.
Own retrieval quality and evaluation loops so agent answers stay grounded — golden sets, regression gates in CI, and honest failure modes.
Role id: applied-ai-evals-rag
Design and ship the coordination layer for multi-agent systems: typed messaging, durable workflows, backpressure, and rollback paths ops teams trust.
Role id: senior-agent-orchestration
Build the consoles and operator UX where humans gate irreversible effects — clear status, audit trails, and calm incident UX.
Role id: staff-full-stack-agent-ops
Send a short note on a system you shipped (or kept alive), a link to work you are proud of, and the role id from the list above.
Fixed-price discovery in 2 weeks. You leave with an architecture, a working spike, and a build plan.