Headway Senior Software Engineer, Agentic Engineering at Headway builds AI-powered internal tools and developer tools to improve mental healthcare software. The role involves coding, impact measurement, and team operations.
Responsibilities
Help Headway's engineers go fast and have a blast as AI agents write more of the code. Find what slows product teams down, and build the agentic tools and habits that fix it.
The fix can be an agent workflow for a manual task that a team does every week. It can be context that teaches agents one part of our codebase. Sometimes a team only needs a tool that we already built. Then your job is to show the tool to them and help them use it.
Some of the work is building. Some of it is teaching: demos, show-and-tells, and short guides that other teams can use. We measure your work by how much faster the teams around you ship, not by how much code you write. You will also help decide what our team builds next.
Product team partnerships. Work with product teams to find the delays that cost the business the most. Remove those delays with agents, tools, or new ways of working. Test the result with the team, collect feedback, and improve it.
Headway-specific agentic tools. Build agent workflows and tools for Headway's code and domains. Examples are context that teaches agents our conventions, automation that removes manual work, and checks that keep quality high as code volume increases.
Adoption and shared defaults. Turn what works into defaults that every team can use, such as starter kits, agent skills, and guides. Share demos, open betas, and short updates. Follow up until teams use them.
Code understanding. Help engineers understand code that agents write. Find new ways to explain, review, and show changes.
Impact measurement. Connect each tool to a business result. Measure throughput, time saved, and adoption. Stop work that does not show results.
Team operations. Share on-call, code review, design review, and interviews with the team.
Cursor, Claude Code, and other AI coding agents · Eddy (our internal AI platform) · Python 3, FastAPI, SQLAlchemy, Temporal, Celery · TypeScript, React · Postgres, Redis · AWS, Terraform · Datadog · GitHub, Buildkite
Qualification
Experience we're seekingYou built internal or developer toolsAgent frameworksContext and retrieval systemsCI/CD and test infrastructureTeaching or enablementYou'll love this role if you want to
Required
You are a senior engineer who uses AI to build every day. You use tools such as Claude Code or Cursor in your own work, and you made your own workflow better with them. For every project, you first ask who the customer is and what problem you solve for them. Your users are other engineers, and you go to them. You do not hesitate to message an engineer that you have never met.
You are comfortable when the best answer changes every few months. When a better tool becomes available, you replace your own tool with it.
Experience we're seeking
You built internal or developer tools, and you got an organization to adopt them. You can explain how you got that adoption and how you measured it.
You built something real with LLMs or coding agents. You know where they fail, and you used evals, guardrails, or human checks to control those failures.
You write strong backend code in a large production codebase. Python is a plus.
You changed how a team works without being its manager. You did this through close partnership, trust, and changes based on feedback.
Preferred
Agent frameworks, workflow engines, or orchestration (Temporal or similar)
Context and retrieval systems, or LLM evals
CI/CD and test infrastructure
Teaching or enablement, such as workshops, internal guides, or champions programs
Experience in healthcare or another regulated domain
You'll love this role if you want to
Help decide how a full engineering organization builds software with agents
Work each week with the engineers who use what you build