Senior Software Engineer, Bite Engineering

• A love of learning a new domain, and shipping something quickly that actually fixes what was broken
• A track record of shipping and owning production systems end to end. Years matter less to us than evidence you've done it
• Strong backend experience, ideally Python in a production context, and comfortable working in a React/TypeScript frontend
• Experience designing and evolving APIs in production. The async tooling behind ours, queues and workers, is something you'll pick up here
• Experience shipping LLM or ML features to production, and measuring whether they actually got better
• Genuine product sense so you can independently weigh technical trade-offs against customer value
• The instinct to keep things simple under pressure, and the judgement to know when a tactical shortcut is right and when it isn't
• Clear communication across deeply technical, semi-technical and plain-language conversations. Contractors buy outputs they can audit, so explaining the working matters
• Don't rule yourself out over the stack. We're looking for a strong engineer with sound principles. Technologies are tools

Desirable

• We're a well-funded startup using AI to decrease risk and grow margin in the architecture, engineering and construction industry
• Our current focus is general contractors. When they bid to build a major project, such as a skyscraper or an airport, they price millions of pounds' worth of project risks, and treatments for them, by hand, and often inaccurately
• During delivery, risk treatments that were suboptimal or never applied erode the margin
• Despite billions in annual revenue, contractors only keep ~2%
• Bite puts contractors' data to work at tender and through delivery, so the bid price rests on evidence and the margin predicted at tender is the margin realised
• Bite ingests the whole tender pack, auto-builds the risk register for the team to review, and prices it with Monte Carlo simulation reconciled against a reference class of similar completed projects. Through delivery, an AI auditor tracks risks and treatments, and every completed project improves the next tender
• You'll be one of our first engineering hires, working directly with the founders and with Phil Cowans, our fractional CTO, who co-founded Songkick and Nested as CTO and studied Bayesian machine learning at Cambridge (the maths behind probabilistic risk pricing). The platform is in production on a Python backend and a React/TypeScript frontend on Azure. You don't need experience in construction or in our stack. We hire for strong core engineering, and the team treat specific technologies as tools for tackling problems
• The problem is genuinely hard. A tender pack is thousands of pages of drawings, specifications, contracts and bills of quantities, with a tight deadline for bids. Turning it into a structured risk register, then into a defensible bid price checked against similar completed internal and external projects, is the core of our product. Every number has to show its working, because contractors won't trust things they can't interrogate and deeply understand
• You'll own whole parts of the product, not a slice of them. There's no ticket factory and no waiting on three other teams. You'll take features from a conversation with a customer through implementation, tests and deployment, and own significant parts of the codebase outright, including the extraction pipeline the risk model depends on
• You'll build it in the room with the customer. We're co-developing the product on real data with some of the world's largest general contractors. You'll hear what works and what doesn't directly from bid and risk teams
• Ship features end to end across the Python backend, the React frontend, and the tests that prove they work
• Build the pipeline that turns tender packs into structured, citable risk data: ingestion, LLM extraction, and the checks that catch what it misses
• Productionise the risk model with Phil, with evals against completed projects that show whether a change made it better
• Build the interfaces where bid and risk teams review the auto-generated register and see the working behind every number
• Design and evolve the contracts between our services, and the async queues and workers behind them
• Make the pragmatic calls on tech debt with the team: what we fix now, what we log, and why