Founding ML Engineer in London

• Location London, in person San Francisco, in person
• Commitment Full-time

Compensation Salary + equity

Roles

Every application is read and kept on file even if now isn't the right time.

Right hand to two founders building a hardware, software and media company.

You run the company behind the company, and you sit close enough to the founders to learn how every decision gets made.

• You may own
• - Money: invoices, bills and the payment run, ready for a founder to approve.
• - The books with our accountants, monthly and quarterly.
• - Payroll, benefits and people admin, first form to last payslip.
• - The dates that keep us legal: filings, renewals, insurance, right-to-work. You own the dates and the evidence; the director signs.
• - The office: building, people, deliveries, kit, health and safety.
• - Onboarding and offboarding: working on day one, nothing left open at the end.
• - Contracts and suppliers: a register that matches what was signed, renewals flagged early, suppliers checked before payment.
• - The founders' inboxes and calendars when they are underwater: triage, drafts, follow-ups, prep, travel.
• - Customer success, from order to device on: set-up, first weeks, a lost delivery, a return, a question that needs a human.

Our voice, and what they tell us comes back to the team.

• - The automation of all of the above: agents and pipelines that chase invoices, reconcile the books, draft replies and watch the dates, so the manual job shrinks every month.
• - Whatever we have not thought of yet. There is no ops department, you are it, and the remit grows as fast as you can carry it.
• You’ll fit if
• + You have run ops, finance admin or a chief-of-staff desk at a small company, or come from consulting, banking or a fund and want the other side.

You liked the parts others avoid.

• + Exact with money and dates. A wrong bank detail or a missed filing is not small to you.
• + Short, clear emails. You keep the thread.
• + AI tools, properly. Claude Code is your working day; you automate a chore rather than repeat it.
• + You want to be in the room where it is decided, not in a back office.
• + You learn by proximity: two technical founders, their calendars, inboxes, mistakes and investors, and you want to see all of it.
• + Good with people, not just paperwork. A customer, a landlord, a creator and an accountant get the same straight, warm version of you.

Payroll or bookkeeping exposure Worked with visas, sponsorship or right-to-work Run an office move, a lease or a supplier dispute Organised events or a community Clinical, science or health background on the side Built or sold something of your own Built an agent or pipeline that stuck

You turn raw acoustic and time-series signals from the body into things people can read and act on. This is the hard, unsolved centre of the company.

• You may own
• - The full modelling pipeline: noisy, unstructured real-world sensor data in, actionable insight out.
• - Signal processing on continuous data: extracting clean, robust features from continuous time-series and acoustic streams (DSP, filtering, audio embeddings).
• - Edge-to-cloud architecture: models that run on-device (resource-constrained audio and sensor processing) and on the backend, plus the judgement of what goes where.
• - Agent context and memory: building the harness that feeds our agent the right things at the right time. Good context and memory are most of the product.
• - Ground truth validation: how we validate that what we tell people is actually true. Rigour is the product.
• You'll fit if
• + Minimum MSc (Ph D preferred) in Computer Science, EE, Acoustics, ML or a related technical field.
• + Strong proficiency in Python and deep learning frameworks (Py Torch or JAX).
• + Deep expertise in acoustic signal processing (spectrograms, audio transformers) and time-series analysis (continuous physiological data, sequence modelling).
• + Experience deploying ML models on resource-constrained edge devices (C/C++, Tiny ML, RTOS or mobile neural engines).

Publications at ICASSP, INTERSPEECH, Neur IPS or ICLR Physiological acoustics (heart or lung sounds) or wearable sensor fusion (IMWUT, Ubi Comp) Bridging raw sensor signals to LLM and agent architectures

How we work

We work a lot.

Weekends will go to Suna.

We are flexible when life gets in the way, because we are only human, but we are here because we are passionate and we want to win.

This will never be a 9-to-5, and we would not want it to be. #Work Hard Play Hard

Goes straight to the founders.

How we handle your application is set out in our Privacy Policy .

We keep your application so we can consider you for this and future roles, and you can ask us to delete yours any time.