Senior Software Engineer - Dev Tools

Full-time · Engineering · San Francisco, CA (3 days/week in office)

A dev tools engineer to bring agentic workflows into our development stream - code review, investigation and debugging, refactoring, testing - measured on delivery metrics, not demos.

This role is different from our others: your customers are the engineers here, and the product is how fast and how well they ship.

We believe agentic workflows belong inside the development stream itself, not just in the editor. Code review, investigation and debugging, refactors, test generation and maintenance, and answering the engineering team's day-to-day questions are all loops that can be built, run, and improved. You'll own building them - and owning them means the unglamorous parts too: keeping them reliable, knowing when one is doing more harm than good, and retiring the ones that don't earn their place.

The bar we hold this work to is evidence. We want measurable movement in how this team delivers software - cycle time, review latency, change failure rate, time to recover, build and test reliability, time lost to environment and CI problems. Part of the job is instrumenting the development pipeline well enough to know where the time actually goes, establishing the baseline before changing anything, and being honest about what moved and what didn't. Enthusiasm for AI tooling is necessary here. It isn't sufficient, and it isn't the deliverable.

You'll work closely with our infrastructure team to keep CI fast and dependable, since none of the rest of it matters if the pipeline is red or slow.

What You'll Do

• Build and operate agentic loops across the development stream - code review, investigation and debugging, refactoring, test generation and maintenance, and an engineering help desk that answers questions about our systems
• Instrument the development pipeline end to end, so we have real data on where engineering time goes rather than anecdotes
• Establish baselines and own the metrics that define success in this role: cycle time, review latency, change failure rate, time to recover, build and test reliability
• Evaluate new tools and models against our actual codebase and workflows - run the comparison, measure the result, and make the call to adopt, park, or drop
• Partner with our infrastructure team on CI: build and test times, flakiness, capacity, and the reliability of the pipeline everyone depends on
• Build the guardrails that let agents operate safely in a regulated codebase - permissions and least privilege, secret handling, provenance and review requirements for generated changes, and an audit trail of what was automated
• Improve the underlying substrate agents depend on: repository structure and conventions, test coverage and speed, internal documentation, local and ephemeral development environments
• Work within the tooling and workflows the team already has - support them, extend them, and make them better, rather than replacing them by default; when a replacement is genuinely warranted, own the case for it and the migration that follows
• Work directly with the engineering team to find where they're actually losing time, and to make sure what you build gets used
• Write production code in Python, and in TypeScript where our application codebase requires it
• Participate in business-hours on-call rotation shared across the engineering team (no nights/weekends)

How We Work

We don't have product managers. You'll find the problems yourself by watching how this team works, decide what's worth building, and defend the tradeoffs - that discovery is part of the job.

This is a build-and-measure role, not an evaluation role. We expect strong opinions about AI tooling and we expect them to be tested: shipped into the workflow, instrumented, and kept or killed on the evidence. A loop that looks impressive in a demo and quietly wastes reviewer attention is a failure, and part of your job is being the person willing to say so.

You're also joining a team that already has tools and habits it relies on. Those are the starting conditions, not obstacles. We're looking for someone who can make the existing environment work better before proposing to change it, and who understands that switching costs are real and fall on other people.

Qualifications

Required

• 10+ years of professional software engineering experience, including substantial time writing production application code - you have to be credible to the engineers you're building for
• Experience building internal developer tooling, developer platform, or build and release systems where other engineers were the users
• Hands-on experience building with LLMs and agentic tooling beyond everyday assisted coding - orchestration, tool and context design, prompt and workflow iteration, evaluating output quality systematically
• Strong CI/CD depth: pipeline design, build and test performance, flaky test diagnosis, artifact and dependency management
• Demonstrated use of measurement to drive engineering outcomes - instrumenting a process, establishing a baseline, and showing what changed
• Strong Python, and the ability to work in a TypeScript codebase
• Ability to track a field that is moving quickly and separate what is genuinely useful from what is merely new
• Exceptional communication and influence without authority - adoption is voluntary, and tooling nobody uses is worth nothing
• Track record of improving an engineering environment you inherited, meeting a team where its tooling and workflow already are
• Comfort with ambiguity and a habit of teaching yourself the domain
• Low-ego approach and a high bar for precision
• Based in the SF Bay Area with the ability to work in-office 3 days per week

What Makes You Stand Out

• Experience building evaluation harnesses or benchmarks for LLM output, especially against a real codebase
• Experience authoring MCP servers or designing tool interfaces for agents
• Experience running a developer productivity or platform function, including DORA or similar metrics programs
• Large-scale automated refactoring experience - codemods, AST tooling, static analysis
• Test infrastructure depth: flake reduction, test selection, parallelization, coverage strategy
• Security or compliance awareness around AI in the SDLC - code provenance, data handling, third-party model risk
• Startup experience on small engineering teams (<10 people)

Technical Environment

• Backend: Python
• Frontend: React/TypeScript
• Database: PostgreSQL
• CI/CD: GitHub Actions
• Cloud: AWS

Why Join Us?

• Leverage: Your work compounds across everything the engineering team ships
• A real mandate: This is a funded, owned function, not a side project someone runs between features
• Ahead of the curve: Agentic development workflows are being figured out right now, and you'd be doing that work rather than reading about it
• Direct Influence: Small team, flat organization, direct access to leadership
• Unique Domain: Support engineers building first-of-their-kind Bitcoin-denominated financial products
• Benefits:
  • Fully covered health, dental, and vision for you and your family
  • Meals provided in office
  • Two annual company off sites (fully paid)

Our Values

• Keep your word. Honor commitments to clients and each other.
• Build for the future. We measure success in decades; look ahead without losing sight of what matters today.
• Be curious. Our advantage is understanding the world, the market, and each other.
• Have fun. Enjoy the work. Find humor where you can. Rest when needed. Say when you need help.

How to Apply

Email careers@meanwhile.bm with your resume and the role you'd be interested in.

Company & Funding

We're building the world's largest long-term insurer, using digital money and AI to serve billions of people profitably. We want anyone, anywhere, to be able to save for their future, protect their family, and build wealth across generations.

We face a once-in-a-century opportunity to build a vertically integrated life (re)insurer. Our product offerings are globally unique, making it possible for us to scale our balance sheet, build modern systems from scratch, and then directly compete to win a market that is 3% of global GDP.

We've raised over $140M to date. Sam Altman and Lachy Groom led our initial raise, and they've since been joined by leading investors in AI, insurance, and Bitcoin, including Northwestern Mutual, Apollo, Bain Capital, Pantera, Haun, Framework, Fulgur Ventures, MS&AD, Mouro, Stillmark, and Wences Casares.

Our Bermuda operating subsidiary holds the the world's only license to issue life insurance denominated entirely in Bitcoin. It's also the only company in the world with audited financials stated in Bitcoin. (If you join us, you can expect to do a lot of things no one's ever done before.)

Engineering at Meanwhile

With the advent of ubiquitous AI tooling, the dynamic range in individual engineering effectiveness is only widening.

At Meanwhile, we're planning for a world in which small, tight-knit engineering teams (supported by a small, tight-knit platform team) own entire lines of business, and are compensated accordingly.

We're attacking a huge market with the leanest, most effective team in insurance. Where incumbents employ a thousand people, we think we can make it work with a hundred or a dozen.

We're looking for hungry ICs (and former managers who see the writing on the wall) interested in pushing the boundaries of engineering productivity in a vertically integrated, regulated organization.

We're growing quickly. You will fix million-dollar bugs.

On the business side, we hire for deep domain expertise, ambition, and the creativity to figure out the previously impossible. Because our engineers work closely with people from the business, they need curiosity, flexibility, an appetite for (and the ability to digest) complex context, and strong communication skills.

Our view is that ownership is taken, not given. You will be successful here if your work progressively builds others' trust in your ability to identify, attack, and solve larger and larger problems, including those that no one else has anticipated.

We believe that "code wins arguments," that prototyping is often the best first step in a design process, and that the impact of velocity is non-linear.

You are excited by putting up multiple meaningful changes, or writing thousands of lines of code, in a day (even though you know that deleting lines of code is more exciting than writing them, and that, occasionally, deep reflection is required in order to ship anything).

You aren't fazed by building systems that don't work out — sometimes, you have to throw code away.

Intellectual honesty is non-negotiable. You love to learn and to teach, to ask questions and to answer them, and to be transparent about your uncertainty.

You are eager to learn, with the rest of the team, how to work with AI tools, including agents, in order to move faster and ship better, more complete versions of your ideas. You will experiment with new ways of working, with the expectation that some of them will be unsuccessful, and you will teach others what works.

This is the most exciting time in decades to be a competent, technically ambitious engineer. We want to offer you the opportunity to see what's really possible and how much better you can get at your craft.