Staff Software Engineer - Experimentation

Building a new mental healthcare system at Headway is only possible because of the scale and leverage that software can provide. The engineering team at Headway is a small but mighty team using technology to build that future (and have a fun time while doing it!).

Principles that guide us:

Focus on the mission: We view software as a means of effecting change in the world, not as an end unto itself. We write software to empower our patients and therapists to better solve the problems they're facing.

Ship small, learn fast: We are building new and novel products and believe that we learn what our users need by quickly shipping and iteration

Everything is a product: Whether it's our patient search experience or a developer productivity improvement, we treat everything we build at Headway as a product with end-users in mind.

Optimize for trust: We believe that engineers should be continually learning. To learn effectively and to be productive, engineers must feel safe asking questions and discussing mistakes.

Tools we use:

Languages: Python 3, TypeScript

Libraries & Frameworks: FastAPI, SQLAlchemy, React, Remix

Experimentation & Analytics: LaunchDarkly, Statsig, dbt, Snowflake

Datastores: Postgres, Redis

Infrastructure: AWS (Fargate, ECS, S3, and more), Cloudflare, Spark and Kafka

Monitoring: Datadog, PagerDuty, Sentry

Version Control: Github

We are looking for a Staff Software Engineer to own how Headway runs experiments. "Ship small, learn fast" is the second principle on the list above, and experimentation is how we actually do it: eleven teams have run hundreds of experiments deciding how patients find a therapist, how quickly new providers get up and running, whether our AI products are working, and how we collect the clinical outcomes data that proves care is effective. You will own the platform all of that runs on.

This is a hands-on technical leadership role with unusually wide reach. You will design and build the core primitives, including how users are assigned to experiments, how we record what someone actually saw, and how we recognize the same person across devices and between browsing anonymously and signing in. You will decide what our long-term experimentation and feature flagging stack should be; we run LaunchDarkly and Statsig today, and that architecture is yours to evaluate and consolidate. You will also make AI a real part of the workflow, from validating an experiment's setup before it starts to catching a broken experiment automatically instead of discovering it in a Slack thread a week later.

You will also evolve our experimentation capabilities to expand into market-based testing. Headway is a multi-sided marketplace, so we frequently need to change something for therapists and measure what happens to patients, which changes both how a test must be designed and how it can be analyzed. We need to test whole geographic markets, not just individual users. We need to hold groups out for long stretches to understand real long-term impact. These capabilities are well understood in the industry and we have not built them yet, and they are the ones our business needs next.

The job is a platform and a program in equal measure. Half of it is the technical foundation. The other half is raising how well the whole company experiments: the defaults, the review, the documentation, and the education that let a team run a trustworthy experiment without needing you in the room. You will gather needs from teams who all want something different, sequence them into a roadmap, and say no clearly, which means acting as part technical product manager for your own platform.

You will sit in the Patient org, where the highest volume of experiments happens, and serve the whole company. Trustworthy results are shared ownership. You will own assignment, exposure, and the experiment data pipelines, which are the parts of the chain that are engineering problems. You will work closely with Data Science, who own statistical methodology, and with the teams who own our event tracking and metric definitions, because a number is only as trustworthy as the weakest link in that chain.

We are looking for a Staff level engineer who wants their impact measured by what other teams can now do. You think in systems, you are uncomfortable with numbers you cannot explain, and you raise the bar for the people around you.

You have owned an experimentation or feature flagging stack end to end rather than only used one. Whether you built it from first principles or built the missing pieces on top of a vendor matters far less to us than having been accountable for whether the results were correct.

You have built the parts that do not come out of the box: assignment that persists correctly, exposure that reflects what a user actually saw, identity resolved across sessions and devices, or the diagnostics that catch a broken experiment early.

You are fluent enough in the statistics to be a real partner to data scientists. You can tell the difference between a result that is weak and a result that is invalid, and you can hold your position on a measurement question with someone who does this full time.

You have debugged the hard class of measurement bugs: traffic that split unevenly, users who ended up in both arms, effects that vanished once you looked at who was actually exposed, identity that broke between anonymous and logged-in sessions.

You have built a program and not only a platform. You have measurably raised how well an organization experiments through defaults, templates, review, and teaching, not only through code.

You have made a build versus buy call on a platform and lived with it. You can reason about vendor risk and about where a warehouse-native approach beats a hosted one.

You have landed a migration on teams that did not ask for one, by making the new path easier than the old path rather than by mandate.

You use AI as genuine leverage in your work and you have clear opinions about where it must not be trusted without a human.

You have driven technical direction beyond your own team through design documents, reviews, and planning cycles, and you regularly mentor senior engineers.

You are willing to flex across the stack. This role touches backend services, front end rendering, CDN behavior, and the data warehouse.

Experience with multi-sided marketplaces, growth or search and ranking systems, or healthcare is a plus, though none is required.