Build and ship the production agent platform: event-triggered queue, model-agnostic runtime, durable state, human review gate, rollback, and full run logging.
Own the evaluation layer: golden test suites, behavioral assertions, a written judge rubric, safety cases, and a CI gate that blocks regressions.
Build and register a portfolio of agents for reporting, drafting, linting, triage, and Q&A, plus a meta-agent that monitors and improves the platform.
Enforce governance in code: risk tiers, least-privilege credentials, tool-permission gates, decision audit logs, and autonomy classes.
Design how the system contacts people, including per-person interruption limits and message bundling.
Own platform tooling: the compiler and validator, inventory integrity, and context and capability distribution.
Compute operating metrics, including a maturity grade per team, from production data.
Instrument the platform's return by logging the work each agent absorbs and computing a monthly value figure.
Requirements
Have shipped production LLM agent systems that others depended on, with real users.
Design formal evaluations, including golden sets, behavioral assertions, judge rubrics, pass thresholds, and CI gating.
Deep API work and authored MCP servers, with knowledge of their failure modes.
Own infrastructure end to end in Python on GCP, with a cloud data warehouse and infrastructure as code.
Strong judgment about how software contacts people and how often.
Strong signal: public work such as an open-source agent framework, MCP server, or evaluation harness.
Strong signal: experience with LLM observability and cost instrumentation.
Strong signal: experience building an internal platform that non-engineers adopted voluntarily.
Done for you
We'll apply to this for you
Denidel's HR team submits and tracks the application on your behalf.
