Agent Engineer [IC4]

Responsibilities

• Design and harden multi-step, tool-using agent loops for agentic experiences, turning research into reliable, observable, and affordable products at enterprise scale.
• Craft pragmatic evaluations using targeted smoke tests, metrics, and dashboards to measure changes in agentic products, avoiding noise while moving fast.
• Decide on model selection, drive upgrades, and fine-tune or distill models to optimize for cost and latency.
• Improve retrieval, ranking, context engineering, and citations to ground models in customer code for more accurate and verifiable answers.
• Treat cost and latency as product features: profile, distill, cache, and right-size models to ship ambitious features sustainably.
• Own agentic slices of the product end-to-end, from problem-framing through rollout and measurement.
• Establish responsible shipping practices for model and prompt changes with evals, dashboards, and guardrails.
• Up-level teammates in agent engineering through pairing, code review, and modeling best practices.
• Set technical direction for the team on agents, models, evals, and roadmaps, backed by evidence.
• Participate in on-call support rotation.

Requirements

• Senior engineer and technical leader with hard-won skills in agent engineering (blend of software engineering, machine learning, and statistics).
• Built, trained, evaluated, and operated models in production.
• Designed multi-step agentic systems that are reliable, observable, and cost-bounded.
• Strong judgment on evaluations, datasets, baselines, metrics, error analysis; knows when changes actually improve things.
• Pragmatic about evaluations and conscious of cost/latency as product constraints.
• Operates autonomously on ambiguous problems: turns rough ideas and customer feedback into plans, prototypes, milestones, and tradeoffs.
• Operates at staff scope: owns highest-risk/ambiguous problems, sets standards, influences beyond immediate team, translates between engineering and business goals.
• Leads through technical excellence and mentorship.
• Real agent engineering depth with point of view on where agents shine vs. deterministic code or human judgment.

Nice-to-Haves

• Experience with code intelligence, large codebases, or developer tools.
• Fluency in turning natural language into structured queries or building summarization/hover experiences.
• Background in retrieval-augmented generation (RAG), context packing, or citation systems for LLMs.