Frontier AI Engineering

Postman•San Francisco, CA

3d•$300,000 - $400,000•Onsite

About The Position

Postman is building a small Frontier AI Engineering team to work directly with ambitious customers to design, build, and bring AI and agentic systems into production. This hands-on engineering role involves writing code, building prototypes, debugging systems, making architecture decisions, and collaborating closely with customer engineering teams. Work may include agent tool use, MCP integrations, RAG pipelines, evaluation frameworks, model routing, and identity and policy layers that enable agents to act safely. The role spans the AI stack, including models and inference, agents, APIs, tools, context, evaluation, security, observability, and production infrastructure. Engineers bring customer learnings back to Postman, identify recurring problems, and help turn them into reusable product capabilities, reference architectures, workflows, and platform features. The initial team members will also help establish the team's technical bar, architecture, engineering practices, operating model, and culture. Postman is headquartered in San Francisco and has offices in Boston, New York, Austin, Tokyo, London, and Bangalore. The company values in-person collaboration. Roles based in its hubs are in-office five days per week, except Bangalore roles, which are currently in-office three days per week and are expected to transition to five days per week by the end of the year.

Requirements

Strong engineering fundamentals, AI fluency, technical depth, curiosity, judgment, speed, and ownership.
Deep technical ability and enthusiasm for building; able to write code, prototype, integrate, debug, experiment, benchmark, and ship.
Able to work directly with customer engineering teams and reason through architecture as well as debug systems hands-on.
Understanding that production LLM systems require systems engineering, evaluation, reliability, data, APIs, and security, not only prompt writing.
Comfortable taking on hard, ambiguously defined problems and carrying them from architecture through production.

Nice To Haves

Background in applied AI, agentic systems, AI infrastructure, distributed systems, developer platforms, APIs, security, ML infrastructure, inference, or model platforms.

Responsibilities

Build production AI systems using LLMs, agents, MCP, APIs, tool calling, RAG, vector search, structured outputs, multimodal models, and AI workflows.
Design agents that work across tools, APIs, enterprise data, and workflows with appropriate identity, permission, policy, and execution controls.
Experiment with foundation models, model routing, inference, prompt and context engineering, and fine-tuning where appropriate; assess quality, latency, cost, and reliability tradeoffs.
Build evaluation systems for agents and LLM applications, including measures for task success, tool correctness, hallucination detection, regression testing, safety, latency, and cost.
Make AI systems production-ready through observability, tracing, guardrails, fallback strategies, rate limits, retries, caching, model gateways, and reliable infrastructure.
Help customers address retrieval, memory, grounding, context windows, embeddings, vector databases, enterprise data, and delivery of relevant context to models and agents.
Develop controls for agent actions, including authentication, authorization, secrets, API access, policy enforcement, human approval, and auditability.
Own loosely defined problems end to end, from architecture and prototyping through evaluation and production.
Prototype, integrate, debug, experiment, benchmark, and ship code.
Work with AI teams, platform teams, developers, and engineering leaders at customer organizations.
Use customer problems and field learnings to influence Postman's AI platform, APIs, Gateway, identity, agent capabilities, developer experience, and product roadmap.
Turn recurring customer problems into reusable primitives, reference architectures, workflows, and platform capabilities.
Help define the team's technical bar, architecture, engineering practices, operating model, and culture.