Deployed Engineer (Professional Services) - AI Agent Systems

We're looking for a Deployed Engineer to join our Professional Services team, working directly with enterprise customers to build reliable, production agents. You'll translate vague enterprise workflows into concrete software specs and guide engineering teams through the resulting solution, or build it for them.

You might spend a week designing a customer's agent architecture, a few weeks co-building their evaluation pipeline, or a quarter embedded inside their team shipping alongside their engineers. You are someone who's built real AI systems for production and can defend the technical tradeoffs within them.

Key Responsibilities

• Advising: Agent architecture design, evaluation strategy review, and best-practice production guidance.

• Building: Co-build with the customer's engineering team across the full Agent Development Lifecycle (ADLC) in outcome-scoped engagements.

• Embedding: Serve as a deployed engineer inside the customer's team for extended engagements, operating as a de facto member of their org to ship agent systems directly.

• Agent Engineering: ADLC end-to-end, architecture design, orchestration patterns, evaluations, custom conversational UIs, and production deployment.

• Applied AI: Post-training, supervised fine-tuning, harness engineering, trace mining, model selection, and evaluation methodology.

Requirements

• 4+ years of software engineering experience with deep expertise in Python. TypeScript/JavaScript a plus.

• 2+ years of hands-on experience building and shipping production agent systems.

• Strong client-facing communication skills, with the ability to confidently articulate architectural decisions to technical stakeholders (engineers, architects, CTOs).

• Strong experience with LangChain/LangGraph/Deep Agents or comparable frameworks, including multi-agent patterns and state management (short and long-term memory).

• Deep familiarity designing and implementing evaluation methodologies for non-deterministic AI systems.

• Comfortable operating across the full spectrum from advisory to embedded delivery.

• Willing to travel up to 20% of the time.

Nice to Have

• Exposure to dataset curation and post-training techniques (SFT, DPO, RLHF) on open-weight models using tools like Axolotl, Unsloth, Hugging Face transformers, or TRL.

• Experience with trace mining to drive continuous improvement loops.