Vetted on a call by a Wave consultant, not scraped. £120,000 and above.
Runs a team of 7 to 8 data scientists doing operational research at the UK's largest grocery retailer, and built the scheduling platform that moves stock from warehouses into stores. PhD-level optimisation applied practically, using presolvers and rules engines alongside metaheuristics and exact approaches. Python and Java, at national scale, in production.
Looking forWants out of retail and into somewhere optimisation is the product rather than a supporting function. Life sciences automation is the domain that came up unprompted. 3 months notice.
How they came acrossComfortable in a scrappy scaling environment, which is not always true of people who have only known one very large employer. Heavy stakeholder management is most of the current job and they handle it without complaint.
Python, Java, git · Metaheuristics, exact optimisation, custom heuristics · Scheduling, logistics, large-scale optimisation
2 employers in 7 years, a research role at a French engineering school then a straight climb from data scientist to lead at the UK's largest grocery retailer. Average tenure 2 years 1 month, longest stint 6 years.
3 months notice, and retail is the only industry on the CV, so a client outside it needs to be relaxed about domain transfer.
PhD in Computer Science.
Seven years at an enterprise data science platform vendor, the last five running professional services for Northern Europe with 9 people reporting in. The work is AI delivery into pharma, defence and financial services, scoping use cases with C-suite sponsors and getting them to production. Python, Spark, Scala, LangChain and HuggingFace, including LLM fine-tuning and prompt engineering.
Looking forWants off the management track and back to building. Specifically wants to scope and deliver AI use cases themselves instead of reviewing other people's. 1 month notice.
How they came acrossComfortable in front of a C-suite and equally comfortable in the code, which is rarer than it sounds. Got to the final stage of a process at a frontier AI lab recently, so they hold up under a hard technical bar.
Python, SQL, Scala · Pandas, Scikit-Learn, Spark · LangChain, HuggingFace, LLM fine-tuning, prompt engineering · A/B testing, recommender systems
4 employers across 12 years, from a large French bank into a music streaming business, then 7 years at an enterprise data science platform vendor across two titles. Average tenure on the record reads 1 year 9 months, which undercounts the 7-year run.
At 212,500 this sits well above the rest of the band, and they're stepping out of a management job into a hands-on one, so settle the level and the money early.
Masters in Data Science.
Builds ML surrogate models for engineering simulations at a physical AI company, forward-deployed into client engineering teams. Across earlier jobs the work covers reinforcement learning, causal inference, uplift modelling and anomaly detection, with a 24% lift on push notification recommendations and a 50% improvement on an emissions optimisation problem as the hard numbers. Python, PyTorch, Dagster, Airflow, AWS.
Looking forDone with client-facing and forward-deployed work. Wants to own a problem internally, end to end, and stay with it long enough to see it mature rather than handing it over.
How they came acrossTook our call at 8pm, jetlagged, from Australia, and still gave a clear account of what they want and why. Communication rated 10 out of 10, and genuinely curious rather than performing it.
Python, SQL, PyTorch · Dagster, Airflow, ElasticSearch · AWS, GitLab · Causal inference, reinforcement learning, uplift modelling
8 employers in 10 years, including a global financial messaging co-operative, a market data business, a causal AI startup and an insurtech, now a physical AI company. Average tenure 11 months.
Average tenure of 11 months across 8 employers will get asked about, with layoffs at the last two accounting for part of it, and the file carries both immediate availability and a 3 month notice period.
Masters in Financial Engineering.
Tell me which came closest and I recalibrate, then come back with 10 to 15 more inside 24 hours.
Show me the wider bank