Software Engineer, Forward Deployed Agent Builder

Brex Brex · Fintech · United States · Engineering

Software Engineer role focused on building and deploying AI agents to automate and augment internal business functions at Brex. The role involves embedding with teams, understanding workflows, designing and shipping agents, integrating them with internal systems, and defining evaluation frameworks. Requires hands-on experience with LLMs, agent frameworks, and tool-use patterns.

What you'd actually do

  1. Embed with partner teams to deeply understand workflows, including shadowing and learning different job functions.
  2. Scope, design, and deploy agents that take over real workflows across internal orgs.
  3. Integrate agents with internal systems, APIs, and data sources.
  4. Define evaluation frameworks, success metrics, and feedback loops to measure and improve agent performance.
  5. Build shared tooling and contribute to playbooks that accelerate deployments, while owning reliability, adoption, and business impact.

Skills

Required

  • 4+ years in engineering, technical product/program management, applied AI, or a similar technical role
  • Hands-on experience building with LLMs, agent frameworks, and tool-use patterns (e.g., MCP, function calling, RAG) across the full stack (data, APIs, orchestration, product).
  • Experience designing and optimizing SQL and/or NoSQL databases, including data modeling, query performance tuning, and schema design.
  • Strong ability to decompose human workflows into scalable, agentic systems, grounded in a deep curiosity for how people work.
  • High ownership and cross-functional influence, with the ability to quickly build trust across teams and seniority levels.

Nice to have

  • Experience automating back-office functions rather than just customer-facing AI features.
  • Experience with OpenClaw, Hermes, or other general-purpose agent harnesses.

What the JD emphasized

  • track record of shipping AI/automation systems to production
  • Hands-on experience building with LLMs, agent frameworks, and tool-use patterns (e.g., MCP, function calling, RAG) across the full stack (data, APIs, orchestration, product).

Other signals

  • building AI agents to automate and augment internal functions
  • design, build, and ship agents that deliver real outcomes
  • wiring together tools, MCPs, and internal systems into agentic workflows
  • spend most of your time building, shipping, and iterating