Software Engineer, Backend Developer

Type Script Contract (3 months initially) Hybrid (London)

Carpata is building the infrastructure to automate parts procurement across the automotive aftersales industry, a market worth more than $2 trillion globally. The hard part is fitment. Garages, dealers and fleets spend a lot of time working out which part fits which vehicle, and they still order the wrong one too often. Solve that and the rest of the process can be automated.

Give us a registration and a job, and we return the right parts and let you order them from your supplier.

We are a seed-stage company with paying customers in production. The founders have built and sold automotive technology businesses before. We are backed by Concept Ventures, Nosara Capital and angels from Car Wow, Just Park, Autotrader and Bumper.

Your first project is to automate the vehicle data pipeline that keeps our catalogue current. Every vehicle on the UK road has to be matched to the right entry in the catalogue, the source data changes every month, and parts of that process still need an engineer to run them. You will build the pipeline that does it on a schedule, so that our catalogue team reviews what it proposes rather than operating it.

The work is a mix of data engineering, matching records across sources that describe the same vehicle differently, and production engineering on AWS. It is not infrastructure-heavy: the Lambda, Postgres and Terraform foundations are already in place.

Beyond that, the work follows the roadmap. We are building in three areas: catalogue intelligence, which is identifying the right part for a given vehicle and catching errors before customers do; workflow automation, which covers search, ordering and integrations with the systems garages and dealers already use; and supply, connecting more suppliers so customers can order through us whoever they buy from.

Over three months you would work in at least two of these. We are a small team and do not specialise early, so we need someone comfortable moving between backend, data and, when needed, front-end work.

You would get up to speed with our Lead Engineer, who owns the platform architecture, and work day to day with Pete, our CTO, and with our catalogue lead, who brings the automotive knowledge.

Type Script throughout the platform, with some existing data tooling in Python. Postgres. AWS Lambda with Terraform. A GraphQL API. Next.js and React on the front end. Most of our code is written with Claude Code, so engineers spend more of their time on design, on directing the implementation, and on reviewing and testing what comes back. We are still working out the best way to build like this, and we want people who are interested in that question, not just tolerant of the tools.

The language you have written most in matters less than it used to, though Type Script is home.

Most of the hard problems are still open, and there are not many of us to solve them. Whoever joins now will have a large say in what we build and how we build it, including how we work with AI tools.

A software engineer, backend or full stack, who will take on a hard problem and see it through, rather than work through tickets.

• You have turned a process that ran by hand into one that runs without a person in the loop.
• You ship small and often: pull requests merged the week they open, then improved on what breaks.
• You build with AI tools every day, check what they produce, and know where they fall short.
• You back your decisions with numbers and share them when they are bad as well as when they are good.
• You have worked somewhere small, where you had to work out the approach rather than be handed a specification.
• You go for the hardest part of the problem first, not the predictable route.
• You build for a mechanic at 6pm on a Friday, not for a demo.

Experience on data projects helps but is not essential. An interest in cars helps but is not essential either.

• You need the problem fully defined before you start.
• You would rather perfect something than ship it.
• You think "good enough" is acceptable when the customer needs it right.
• You design for the happy path and hope.
• You dismiss feedback that challenges an assumption rather than getting interested in why.

Three months to start, five days a week. If it is working for both sides we would look to extend, or talk about a permanent role. Equity is offered to…