Jobbit Labs is the R&D and real-world-data division of Jobbit, building proprietary datasets and world models for physical, embodied AI. These are full-time, employed roles in London, for engineers, data, design, commercial and operations people who want their work at the foundation of how machines understand the real world.
Roles across engineering, machine learning, data operations, design, business development and office operations. Every role is hands-on, early-stage and close to the work that makes our datasets real. These positions are currently closed, we're not actively hiring right now.
A full-stack software engineer role at an early-stage deep-tech AI startup in London. You will build the Jobbit platform that books and runs real-world tasks, and the internal tooling and pipelines that turn how physical work actually gets done into training, fine-tuning and evaluation data for AI labs and robotics teams. This is hands-on TypeScript/React/Node work on edge infrastructure, with a short path from your commit to production.
What you’ll do
Ship full-stack features across the Jobbit platform (https://jobbit.uk): front end in React/Next.js, back end in Node/TypeScript on Cloudflare's edge
Build and maintain the internal tooling and data pipelines that ingest, clean, label and version real-world task data into datasets
Design and implement APIs and services that move task data reliably from capture through to dataset delivery
Own data plumbing end to end: schemas, validation, storage (D1/KV/R2), queues and scheduled jobs
Write pragmatic tests, instrument what you ship, and keep the system observable in production
Work directly with data operations and ML colleagues to turn messy real-world inputs into clean, queryable datasets
Help drive sign-ups and engagement on jobbit.uk, including the flows around the CV-drafting AI assistant
Review peers' code, keep the codebase lean, and make sensible build-vs-buy calls at startup pace
What we’re looking for
Strong full-stack engineering with TypeScript across the stack, plus solid React (Next.js a bonus)
Comfortable building and consuming production APIs and the data plumbing behind them
Pragmatic about data: schema design, validation, and moving data between services reliably
You write code that is readable, tested where it matters, and observable once it ships
Able to own a feature from idea to production with light supervision
Happy with ambiguity and changing priorities: you scope, ship and iterate rather than wait for a spec
Clear written communication and a collaborative, low-ego working style
Right to work in the UK and able to work hybrid from our London office (124 City Road, EC1V 2NX)
Nice to have
Experience with Cloudflare Workers, Pages, D1, KV, R2 or other edge/serverless platforms
Background in data pipelines, ETL/ELT, or dataset tooling for ML
Exposure to physical/embodied AI, robotics data, or world models
Early employee experience at a startup, or a side project you shipped end to end
Familiarity with queues, workflows and durable/scheduled execution
As a Machine Learning Engineer at Jobbit Labs, you train and evaluate world models on the proprietary real-world, multimodal data we record of physical work actually getting done, then turn it into training, fine-tuning and evaluation signal for AI labs, robotics teams and enterprises. This is a research-leaning role that ships: you will work hands-on in Python and PyTorch on vision-language-action and multimodal models, sim-to-real, and embeddings, and own the data and eval pipelines that prove a model is getting better. A strong fit for anyone searching for machine learning engineer jobs in the UK who wants physical and embodied AI problems at an early-stage deep-tech company, not another wrapper around someone else's API.
What you’ll do
Train, fine-tune and evaluate world models and vision-language-action (VLA) models on our real-world multimodal datasets (video, audio, motion, language)
Build and maintain reproducible data and evaluation pipelines that turn raw recordings of physical work into clean training, fine-tuning and benchmark sets
Design eval suites and metrics that measure whether a model actually understands physical tasks, and make regressions impossible to miss
Run sim-to-real experiments and close the loop between simulated environments and recorded real-world behaviour
Develop and probe embeddings and representations of multimodal physical-work data for retrieval, clustering and downstream training
Profile and optimise PyTorch training runs for throughput, memory and cost on GPU infrastructure
Read the relevant literature, prototype quickly, and turn promising research ideas into production-grade, shippable code
Work closely with the data operations and product teams to specify what new data we should capture and how it should be labelled
Document experiments, datasets and model behaviour clearly so results are trustworthy and repeatable
What we’re looking for
Strong Python and hands-on PyTorch experience training and evaluating deep learning models end to end
Solid grounding in modern ML: transformers, multimodal/vision-language models, embeddings and representation learning
Experience building data and evaluation pipelines: you treat datasets and benchmarks as first-class engineering artefacts, not afterthoughts
Comfort reading research papers and reimplementing or adapting methods into working code
A bias to ship: you can take an idea from notebook to a tested, maintainable training or eval pipeline
Sound engineering fundamentals: version control, testing, reproducibility, clean code
Clear written and verbal communication about experiments, trade-offs and results
Based in the UK and able to work from our London office (124 City Road, EC1V 2NX) on a hybrid basis
Nice to have
Experience with vision-language-action models, robotics learning, or embodied/physical AI
Familiarity with sim-to-real transfer, simulation environments, or imitation/reinforcement learning
Experience handling large-scale video or multimodal datasets and the tooling around them
Distributed/multi-GPU training experience and performance optimisation
A relevant publication, open-source contribution, or strong portfolio in ML; early-career and recent graduate candidates with the right depth are welcome to apply
Jobbit Labs records how real physical work actually gets done and turns it into training, fine-tuning and evaluation data for AI labs and robotics teams. As Data Operations Specialist you own the quality of that real-world dataset end to end: curating, annotating and QA-ing the multimodal task trajectories that make our world models trustworthy. It is a hands-on data operations role at an early-stage deep-tech company in London, sitting at the point where raw recordings of physical work become clean, structured, model-ready data.
What you’ll do
Curate, label and structure multimodal task trajectories (video, audio, sensor and text) captured from real physical work
Run QA passes on annotated data and own the metrics that define what 'high quality' means for our datasets
Design, write and maintain annotation guidelines, taxonomies and labelling schemas, and keep them versioned as the data evolves
Build and operate annotation workflows: defining tasks, routing batches, and managing throughput and turnaround
Spot-check, audit and resolve edge cases, ambiguous labels and inter-annotator disagreements
Work closely with engineers to improve labelling tooling, surface friction and shape the data pipeline
Track data provenance, consent and licensing so every record is clean and compliant
Report on dataset coverage, quality and gaps to the data and research teams
What we’re looking for
Hands-on experience in data operations, data annotation, data QA or a data-labelling/curation role
A sharp eye for detail and a genuine instinct for what makes data clean, consistent and trustworthy
Comfort writing clear annotation guidelines and turning fuzzy real-world cases into precise, repeatable rules
Working knowledge of SQL or spreadsheets for slicing, auditing and reporting on datasets
Experience with annotation or labelling tools (e.g. CVAT, Label Studio, or similar)
Strong written communication and the discipline to document decisions as you go
Comfortable operating with ambiguity and shifting priorities in an early-stage environment
British English fluency and right to work in the UK
Nice to have
Experience labelling multimodal data: video, audio, sensor or pose data
Familiarity with computer vision, robotics or embodied/physical AI datasets
Basic Python for scripting QA checks, transforms or data audits
Experience managing or coordinating a pool of annotators or reviewers
Exposure to data provenance, consent or licensing in a regulated or sensitive-data setting
We are hiring a Product Designer to own end-to-end UX/UI across Jobbit Labs, from the Jobbit platform that books and runs real-world work (jobbit.uk) to the internal data tools our team uses to record, label and evaluate how physical work actually gets done. This is a hands-on product design role at an early-stage deep-tech company building real-world datasets and world models for physical AI. You will turn genuinely complex workflows into clean, usable interfaces, and set the design foundations the rest of the company builds on. One of the best product designer jobs in London for someone who wants real ownership over the craft and the system.
What you’ll do
Own product design end to end: from problem framing and user flows through wireframes, high-fidelity UI and shipped product
Design clean, legible interfaces for genuinely complex workflows: task booking and dispatch on jobbit.uk, plus internal data capture, labelling and evaluation tools
Build and maintain our design system (components, patterns, tokens and documentation) so the product stays consistent as it scales
Prototype quickly to test ideas, then refine to production-ready specs alongside engineering
Run lightweight product research and usability testing with real users: freelancers, customers and our internal data team
Partner closely with engineers and product to ship, reviewing implementation for pixel and interaction fidelity
Design with an instrument-panel sensibility: dense information made calm, precise and easy to act on
Contribute to brand and marketing surfaces that drive traffic and sign-ups to jobbit.uk
Help define the design process and raise the quality bar as one of the first designers in the company
What we’re looking for
A strong portfolio of shipped product work, ideally including complex or data-heavy web applications
End-to-end range: comfortable across UX, interaction and polished visual/UI design
Fluency in a modern design tool (Figma) and experience building or extending a design system
Ability to design for complexity: dashboards, multi-step workflows, tooling and dense data, made simple
Solid grounding in interaction patterns, accessibility and responsive design
Comfort working closely with engineers and translating designs into precise, buildable specs
Self-direction and high standards: you can own ambiguous problems and ship without heavy process
Clear communication and a collaborative, low-ego approach to feedback
Nice to have
Experience designing internal tools, annotation/labelling interfaces or data-ops workflows
Familiarity with HTML/CSS or front-end handoff (React/Tailwind a plus)
Background in marketplaces, two-sided platforms or scheduling/booking products
Interest in AI, robotics, physical/embodied AI or real-world data
Motion and prototyping skills for richer interaction design
Early-stage startup experience and a builder's mindset
You will open and close the deals that put Jobbit Labs' real-world datasets in front of the people building physical and embodied AI. We record how real physical work actually gets done and turn it into training, fine-tuning and evaluation data for AI labs, robotics teams and enterprises, and we need a commercial operator who can sell that to technical buyers. This is a founding-stage business development role at a London deep-tech company: you own pipeline, run outreach, shape data-licensing and partnership offers, and carry deals from first contact to signature.
What you’ll do
Build and own the commercial pipeline for data-licensing and partnership deals with AI labs, robotics teams and enterprises
Run targeted outreach to research leads, applied-AI teams and procurement, and convert cold contact into qualified conversations
Lead deals end to end: discovery, scoping, pricing, negotiation, redlines and close, working with founders on terms
Translate what our real-world datasets and world models do into clear commercial value for technical buyers (ML, robotics and data leaders)
Shape data-licensing offers, partnership structures and pilot-to-contract paths alongside the data and engineering teams
Map target accounts in the physical-AI, embodied-AI and robotics ecosystem and prioritise where to spend time
Keep an accurate, instrument-panel view of pipeline (stages, forecast and next actions) in the CRM
Feed signal from the market back into roadmap, dataset priorities and packaging
Represent Jobbit Labs at industry events, labs and partner meetings, and help drive awareness toward the wider Jobbit platform
What we’re looking for
Proven B2B business development or sales track record closing technical or data/API products, ideally in AI, ML, data, robotics or developer tooling
Comfort selling to technical buyers: you can hold a credible conversation about datasets, models, evaluation and how AI teams actually work
Strong outbound instincts: you can build pipeline from zero and are happy doing the outreach yourself
Skilled at deal-shaping and negotiation, including pricing, licensing terms and commercial structure
Clear, precise communicator who writes well and can make complex offers easy to say yes to
Self-directed and comfortable with ambiguity: you thrive at early stage without a playbook handed to you
Disciplined pipeline and CRM hygiene; you forecast honestly and chase the right deals
Based in or able to work hybrid from London (124 City Road, EC1V 2NX), with some remote flexibility in the UK
Nice to have
Existing network across AI labs, robotics teams or enterprise AI functions
Experience licensing data, IP or training corpora to AI/ML teams
Familiarity with physical AI, embodied AI, robotics or real-world / world-model data
Background spanning a technical and commercial role (e.g. solutions, partnerships or product-adjacent sales)
Early-stage or founding-team experience at a deep-tech or data startup
LocationLondon (124 City Road, EC1V 2NX) · HybridDepartmentOperationsTypeFull-time · Permanent
Jobbit Labs builds proprietary real-world datasets and world models for physical, embodied AI, recording how real work actually gets done and turning it into training and evaluation data for AI labs and robotics teams. As our Operations & Office Administrator, you are the person who keeps an early-stage deep-tech company running: the office, the calendar, the suppliers, and the day-to-day finance and HR admin that lets the engineering, data and commercial teams move fast. If you are organised, proactive and genuinely trustworthy, this is a high-trust operations job at the centre of a London AI startup.
What you’ll do
Run the London office at 124 City Road day to day: supplies, equipment, post, access, deliveries and the small things that keep a team unblocked
Own scheduling and diary coordination: book meetings, manage room and travel logistics, and keep the founders' time protected and tidy
Coordinate suppliers and vendors (software subscriptions, hardware, recording kit, facilities and contractors), chasing quotes, renewals and deliveries
Support basic finance operations: collect and file invoices and receipts, prepare expenses, and liaise with our accountant/bookkeeper so nothing slips
Support HR and people operations: help onboard new starters, manage equipment handovers, track holiday and keep employee records accurate and confidential
Keep documents, trackers and shared drives organised so the team can find what they need without asking
Help plan and run team events, all-hands, on-sites and the occasional data-recording day or off-site
Be a first point of contact for the office: handling queries, visitors and incoming requests calmly and quickly
Spot recurring friction and propose simple process or tooling fixes that save the whole team time
What we’re looking for
Proven experience in office administration, operations, EA/PA or team coordination, ideally in a startup or small fast-moving team
Genuinely organised: you track many small threads at once and nothing falls through the cracks
Proactive and self-directed: you fix things before being asked and are comfortable with ambiguity at an early-stage company
Trustworthy and discreet, with sound judgement around confidential finance, HR and people information
Strong written and verbal communication in British English
Comfortable with everyday tools (calendars, email, spreadsheets, docs and modern SaaS) and quick to learn new ones
Reliable with light finance admin: invoices, receipts, expenses and working with an external accountant
Based in or near London and happy to be in our City Road office regularly (hybrid)
Nice to have
Exposure to basic bookkeeping or finance tools (e.g. Xero, QuickBooks)
Experience supporting HR/people ops or onboarding new hires
Experience coordinating suppliers, facilities or events
Interest in AI, robotics or deep tech: curiosity about what the team is building
Experience using AI assistants to speed up admin and drafting
Applications are closed for the moment. We’re a London deep-tech team building the real-world data behind physical AI, and we keep sharp people in mind, so check back when roles reopen.