Build the system that predicts what retail does next.

We are a team of founders, builders, scientists and engineers. If you are obsessed with how people actually shop, and you would rather model the next range review than report on the last one, you have come to the right place.

0
Behavioural science principles
0M+
Products analysed
0+
Attributes
0
People, and hiring
Tesco, dunnhumby, Ocado, Iceland
01Why this matters

Everyone shops. Retail is the one industry that will never die.

Every person on earth makes hundreds of small commercial decisions a week, and almost none of them are understood by the people selling to them. Culture now moves faster than the systems built to track it.

What retail runs on today

Historical sales data, and instinct. EPOS shows one scenario, the version of events that already happened, so range reviews reward what sold last year and new products are signed off on a hunch. Millions of pounds go into launches every year that were never going to work, and nobody finds out until the shelf tells them.

What we are building instead

A predictive operating system for retail: one that connects every cultural, behavioural and commercial signal, then maps it against every attribute of every product on shelf. That is what lets Breakthrough model the futures EPOS cannot reach, and it is already shaping decisions inside some of the biggest retailers in the country.

02Momentum

Why now is a good time to join.

Trusted by the retailers that set the category

Breakthrough is grounded in 28 behavioural science principles, applied across 4M+ products and 56,000+ attributes, and trusted by Tesco, dunnhumby, Ocado and Iceland. The work goes straight into real range reviews and buyer rooms.

The category is ours to define

There is no established playbook for predictive retail intelligence, because nobody has built one before. What gets decided here in the next year sets the shape of the category.

We are growing

Twenty-two people today, across product, engineering, data and client work, with more roles opening as the system scales into new retailers and new markets.

03Fit

You will like it here if…

You can tell more about a company from what it does on an ordinary Tuesday than from anything written on a wall. This is what ours looks like.

The traits we actually hire for.

You would rather ship than wait for certainty.

We work in a category that did not exist two years ago, so there is rarely a settled answer to copy. The people who do well here make the call with the evidence available, put it in front of a retailer, and correct it quickly when the market says otherwise.

  • Deep in more than one discipline, and able to work across the rest. Not a specialist who only goes deep, and not a generalist who only goes wide.

You care how the work is built.

Our output lands in range reviews and buyer rooms at Tesco, dunnhumby, Ocado and Iceland. A number that cannot be traced back to its source is worse than no number at all, so the standard is high and the detail genuinely matters.

  • Our customers are the buyers and category teams at the retailers. Their decisions are the ones our work has to survive.

You want to build something that outlasts you.

Breakthrough is early, and the decisions made now will still be visible in the product in five years. That is a reason to argue properly, hold a high bar, lift the people around you, and leave the ego at the door.

  • Everyone here is involved in hiring, and everyone is expected to hold the line on it rather than lower it to fill a seat.
04Candour

You will probably not enjoy it here if…

We would rather be straight with you now than have you find this out in month three. Every one of these has another side. Turn them over.

Scroll sideways6 cards

05Who you'd join

The people you would be working with.

Twenty-two of us, across London and beyond. No stock photography, no invented job titles.

Three words for your work life here.

  • Fast, unfinished, ours.Adam
  • Ambitious, chaotic, honest.David
  • Curious, exacting, relentless.Sophie
  • Busy. Rewarding. Loud.Aaron
  • Ambiguity, then clarity.Mateo
  • Build, break, rebuild.Oriol
  • Evidence over opinion.Alexey
  • Steep learning curve.Noah
  • Warm, fast, chaotic.Cami
  • Deep end, immediately.Phil
06Day to day

Life at Really Good Culture.

Five days a week by London Bridge, which mostly means the river at lunchtime, a lot of product samples, and somebody else’s dog under the desk.

Runs along the river, tennis in the park

Really good cats and dogs

Whatever category we are analysing this month

Fun in the office

And out of it

The actual work

07Ownership

What you get by being early.

Real influence, immediately

We are twenty-two people. What you build and argue for now will still be visible in the product in five years, because there is nobody above you quietly rewriting it.

No red tape

You own your remit. There is no approval chain to walk a decision through, which is liberating or terrifying depending on the day.

A category with no playbook

Nobody has built predictive retail intelligence before, so there is no established way of doing this for you to inherit. You get to decide what good looks like.

08The package

Perks and benefits.

What we offer today.

33 days off a year

25 days of holiday, plus the eight UK bank holidays on top.

Unlimited AI access

Every AI tool you need to do the work, covered, and without an approval queue in front of it.

Four months paid parental leave

Four months at full pay if you are the primary carer, and four weeks if you are the secondary carer.

Sabbatical after two years

Once you have been here two years you can ask for a sabbatical. We would much rather you took one and came back than left to get the same thing.

Really Good cats and dogs welcome

Bring yours in. Several people already do, and nobody has complained yet.

The newest tech, and a setup to match

A new machine and whatever you need around it. Some setups have got considerably out of hand.

Unlimited food, snacks and drinks galore

Team breakfast every Monday, tea, coffee and kombucha all week, and a steady supply of whatever category we happen to be analysing this month. A takeaway on us if you are here late.

Monthly socials, and the rest

A social every month, plus the drinks and dinners that happen anyway. Easier when people actually like each other.

Christmas closure

The office shuts over Christmas, and it does not come out of your holiday allowance.

Flexible hours

Flexible on hours, not on days. The day starts at 9:30 and how you shape the rest of it is largely up to you.

London Bridge, and everything round it

The river is a few minutes away for anyone who runs at lunch, the tennis courts in the park get more use than you would expect, and there is far too much food within five minutes of the door.

Cycle to work scheme

A new bike and everything that goes with it, bought through the company and spread across the year, which works out around a third cheaper than buying it yourself. The river path starts a few minutes from the door.

Five days a week, together

All of us in, five days a week, with flexibility on the days you genuinely need it. The people here become your friends, which is the difference between a job you do and a job you actually like.

09Hiring

Open roles.

Three engineering roles, one product role, and one open application. The most urgent need is data and AI engineering: applied statistics and production data infrastructure in the same person. Every role has a full description attached.

Engineering

  • AI Data Engineer

    The most urgent vacancy. Applied statistics and production data infrastructure, in one role.

    LondonFull-time
    Full description

    This role covers the statistical work and the data infrastructure it runs on. The prediction engine spans 4M+ products and 56,000+ attributes. You would own the pipelines that produce and maintain that data, and the scoring and evaluation work built on top of it.

    What you would be doing

    • Design, build and maintain the ingestion, transformation and storage layers behind the prediction engine.
    • Implement the 28 behavioural science principles as features consumed by the scoring models.
    • Build and tune scoring systems, including the Breakthrough and Buzz scores.
    • Define evaluation methodology and baselines, and measure every model or score change against them.
    • Integrate LLM-based extraction into production pipelines, including validation, retry and fallback handling for non-deterministic output.
    • Implement data quality controls: validation at ingest, provenance tracking and confidence scoring.

    Stack

    • Python
    • SQL
    • PostgreSQL
    • Pipeline orchestration
    • LLM APIs
    • TypeScript (optional)

    What we are looking for

    • Applied statistics: distributions, sampling, significance testing and experiment design.
    • Production data engineering: scheduled pipelines that other systems depend on.
    • Experience deploying and maintaining models or scoring systems in production, not only prototyping them.
    • SQL to the level of schema design and query optimisation.
    • Experience working with incomplete or inconsistent source data, and the validation that requires.

    With your applicationDescribe a pipeline or dataset you built that other systems depended on. What were its failure modes, and how did you detect them?

    Apply
  • Full Stack Engineer, back end focused

    A standing vacancy. We hire back end engineers whenever we find strong ones.

    LondonFull-time
    Full description

    The back end covers the extraction and scoring pipelines, the data model they write to, and the APIs the B2B dashboard and the mobile app read from. This role owns that end to end rather than one service within it.

    What you would be doing

    • Own back end features from schema design through to the API consumed by the dashboard and the mobile app.
    • Build and maintain the pipelines that convert shopper transcription into structured behavioural data.
    • Implement the 28-dimension behavioural scoring that produces each reviewer profile.
    • Integrate LLM extraction prompts into the pipeline, with validation and error handling for non-deterministic responses.
    • Design and maintain REST APIs between the database, the B2B dashboard and the mobile app.
    • Profile and optimise query and pipeline performance as data volume grows.

    Stack

    • TypeScript
    • Node.js
    • Next.js
    • PostgreSQL
    • REST APIs
    • Git
    • LLM APIs

    What we are looking for

    • Back end engineering depth: schema design, API design and query optimisation.
    • Working React and Next.js knowledge, sufficient to deliver a feature end to end.
    • Experience with SQL, third-party APIs and inconsistent source data.
    • Able to ship production code with limited supervision and to review other engineers.
    • Experience designing for change: migrations, versioning and backwards compatibility.

    With your applicationDescribe a back end system you built that had to handle unreliable inputs. What failed, and what did you change?

    Apply
  • Full Stack Engineer, product and dashboard

    The B2B dashboard, the internal tooling behind it, and the API layer between them.

    LondonFull-time
    Full description

    This role covers the B2B dashboard that brands and retailers use, the internal tooling Ops, Data and Content run on, and the API layer connecting them. Features are owned end to end, from data model to interface.

    What you would be doing

    • Build and extend the B2B dashboard across data model, API and front end.
    • Develop internal tools that automate Ops, Data and Content workflows.
    • Build and maintain REST APIs connecting the dashboard, the database and the mobile app.
    • Implement new views and data visualisations, and iterate on them against client usage.
    • Integrate retail, web, social and sales data sources into the application layer.

    Stack

    • TypeScript
    • React
    • Next.js
    • Node.js
    • PostgreSQL
    • REST APIs
    • Git

    What we are looking for

    • Production experience with React, Node and PostgreSQL in TypeScript.
    • Front end fundamentals: component architecture, state management and render performance.
    • API design and integration experience across more than one consumer.
    • Able to scope and ship incrementally and to revise against feedback.
    • Able to work from a specification that changes during delivery.

    With your applicationDescribe a feature you shipped incrementally. What did you learn once it was in production that you had not anticipated?

    Apply

Product

  • Product Owner, or a product-driven engineer

    Either a product owner with technical depth, or an engineer who wants to own scope.

    LondonFull-time
    Full description

    This role owns a product area: its roadmap, its specifications, and the measured outcome after release. It suits either a product owner with technical depth or an engineer who wants to own scope rather than tickets.

    What you would be doing

    • Own a product area end to end: roadmap, prioritisation, specification and release.
    • Translate commercial requirements from retailers and brands into technical specifications.
    • Write user stories, flows and edge cases, and maintain them as the product changes.
    • Work directly with engineering and data on scope, sequencing and trade-offs.
    • Define success metrics before release and measure against them afterwards.

    What we are looking for

    • Evidence of owning product outcomes and the decisions behind them.
    • Able to write a specification an engineer can build from without repeated clarification.
    • Technical enough to discuss architecture and data model trade-offs.
    • Experience defining and instrumenting product metrics.
    • Retail or FMCG commercial knowledge is useful but not required.

    With your applicationDescribe a product decision you would now make differently. What information did you have at the time, and what changed your view?

    Apply

Open application

  • Tell us what we are missing

    An open application. We read every one.

    LondonOpen
    Full description

    An open application, read in full. It works best when it is specific about the scope you would take and what you would deliver in it.

    What you would be doing

    • State the area you would own and what would be measurably different in six months.
    • Identify a specific gap or problem in what we currently do.
    • Include work you have delivered, with links where possible.

    What we are looking for

    • A specific proposal rather than general interest.
    • Evidence of relevant delivered work.
    • A response that engages with the detail on this page.

    With your applicationIn no more than 200 words: what scope would you take, and what evidence supports that you can deliver it?

    Apply
Not sure which of these you are? Apply anyway and say so

Apply.

Tell us which role you are interested in and why. We read everything that comes in.