Founding ML Engineer - Tech Lead (CTO path)

Founding ML Engineer - Tech Lead (CTO path)

Full-time · 12-month fixed term (funded) · £45,000 + equity · UK-based, London a bonus · Start 1 September 2026

About us

We're an early-stage consumer health-tech startup in stealth, building for the millions of people with chronic conditions who end up managing their health largely alone once standard care plateaus. We look at health as a system, from the consumer side, and we're building toward two outcomes: better day-to-day health outcomes in chronic disease and, long-term, feeding what we learn into how chronic health is understood, all the way up to policy.

At the core is a hard ML research problem: saying something reliably useful for one individual from limited data, where calibration matters because the stakes are high.

We're funded by a secured, non-dilutive UK innovation grant, won on an extensive evidence base with a waitlist, + national partnership, + a mapped regulatory position, including a 35-page bespoke regulator review + a working, granular, custom automated go-to-market engine. The founder is an engineer + scientist + builder with a lifetime of research into their own disease, after a career in private equity and managing the allocation of $850M/yr.

The role

The next 12 months are a funded build, and this role is its technical core. You are hire #1, working directly with the founder, with a full-stack developer supporting you for the first four months. You own the technical side end to end: the models at the heart of the product and the platform they run on.

Over the year you will:

• Set the technical foundations. Set up the engineering practices you actually want to live with (code review, standups, issue tracking).
• Model the domain. Co-design a proprietary domain ontology with the founder and build the knowledge-graph layer on top of it.
• Build the extraction pipeline. LLM-driven information extraction from free-text and structured user inputs into the knowledge graph, with confidence scoring and human-in-the-loop validation.
• Build the inference layer. Work out which inference approaches - from hierarchical Bayesian models to causal ML and N-of-1 designs - give well-calibrated answers. The deliverables are benchmarks and calibration reports.
• Learn from a live cohort. Real users co-design and use the product from the first months. You'll analyse their data as it accumulates and recalibrate the models continuously; the strongest findings become our first research output.
• Work to a standard that survives scrutiny. We operate deliberately inside a defined pre-regulatory boundary; your methods and evaluation documentation double as the evidence base for the regulated features that may follow.