Skip to content
×

Product Scientist

Job in Greater London, London, Greater London, W1B, England, UK
Listing for: Bolt Insight
Full Time position Listed on 2026-11-18
Job specializations:
Salary/Wage Range or Industry Benchmark: 90000 - 120000 GBP Yearly GBP 90000.00 120000.00 YEAR

Job Description & How to Apply Below

Location: Greater London

About the role

We’re looking for a scientist who treats a model as a product. You want to know who will rely on its output and what decision it supports before choosing an architecture, and you measure success by whether researchers trust and use the result, not by whether the model shipped.

At Bolt Insight, our platform helps product, marketing and research teams understand their customers through AI-moderated interviews, real-time chat, media analysis and research insights. The Simulation Suite is our next big step: AI Personas, Digital Twins and synthetic respondents that let teams explore consumer behaviour faster, at lower cost and at a scale real fieldwork can’t reach.

In market research, a synthetic respondent is only useful if the industry can trust it, so every release passes a methodology gate before it reaches customers. As Product Scientist, you decide what the Simulation Suite should become scientifically, and you build and validate it yourself. You’ll own the technical direction and roadmap together with Product, Research and Engineering, and spend most of your time hand-on with data and models.

What you’ll be doing
  • Build and prototype the models. Persona and twin construction, calibration against real survey and interview data, prompt and agent design, retrieval and evaluation pipelines. You’ll write and review code and pair with engineers on the path to production.

  • Define and run validation. Design the evidence we need to prove a twin or persona is good enough: benchmarks against ground-truth studies, distributional fidelity, consistency, bias and drift checks. Build our synthetic validation framework and an evidence pack aligned with ESOMAR and MRS guidance, and work with our methodology team to pass the release gate.

  • Own the Simulation Suite roadmap. Turn bets into clear, shippable increments: category-specific Digital Twins, the Digital Twins MVP for pilot clients, agentic quantitative segments, always-on synthetic panels and moderated focus groups, and eventually multi-category, self-serve twin creation. Prioritise, write crisp specs and keep the work moving.

  • Make the trade-offs explicit. Decide what is good enough for a pilot versus GA, and where synthetic data complements real fieldwork and where it shouldn’t replace it. Communicate those limits honestly to customers and internal teams.

  • Work with LLMs in production. Build on OpenAI, Claude and Gemini using Lang Chain and Lang Graph, and evaluate with Langfuse. Balance quality, latency and cost, and design evals that catch regressions before customers do.

  • Work with pilot clients. Join discovery with pilot customers, understand the research questions they want to answer, and bring what you learn back into the roadmap.

  • Measure what you ship. Define success metrics before you build, track adoption and trust signals with Post Hog, and use the results to decide what to improve next.

About you
  • 5+ years in data science, applied ML or ML engineering, with models you’ve taken to production and can talk about in terms of impact.

  • Strong Python and solid applied statistics: sampling, weighting, calibration, uncertainty and experimental design.

  • Hands-on experience building LLM-powered systems (prompting, retrieval, agents, evaluation), and a clear view of where LLMs are reliable and where they aren’t.

  • Experience validating model output against ground truth, and the rigour to say “this isn’t ready” when the evidence isn’t there.

  • A track record of shaping what gets built, not only how: you can take an ambiguous problem, ask the right questions and turn it into a prioritised plan.

  • A great communicator who can explain statistical and modelling trade‑offs in plain language to researchers, product managers and clients.

  • Comfortable reading…

To View & Apply for jobs on this site that accept applications from your location or country, tap the button below to make a Search.
(If this job is in fact in your jurisdiction, then you may be using a Proxy or VPN to access this site, and to progress further, you should change your connectivity to another mobile device or PC).
View / Apply for Jobs
Matching My Jurisdiction

Search for further Jobs Here:

SEARCH JOBS: 10+ Million Listings
(Try combinations for better Results! Or enter less keywords for broader Results)
Education Level
Experience Level (years)
Posted in last:
Salary