Principal Data Scientist

My client is a venture-backed data and analytics business. They take large, messy real-world datasets, run their own prediction models over them, and sell the output as a SaaS product to professional users around the world who rely on it for time-sensitive decisions.They’re hiring a Principal Data Scientist to be the senior data science authority in the business. It’s a hands-on individual contributor role with no direct reports, reporting into a VP-level data leader. A lot of the work is on problems nobody has solved yet, where there’s no existing baseline and you’d be the one deciding what good looks like.What you’d be doingSetting data science direction at a product level, deciding what gets built and whyOwning models end to end, from research and experiment design through to production, monitoring and long-term maintenanceWorking mostly on time-series and anomaly detection, alongside sequence-to-sequence models and stochastic optimisationImproving the quality of models that are already liveLeading the company’s Data Science Guild, mentoring and reviewing work across the practiceExplaining and defending your models to traders, analysts and non-technical colleaguesWhat you’ll needExperience operating at Staff or Principal level, with the autonomy that comes with itDeep grounding in ML theory and the maths behind itFluent, production-standard PythonA track record of putting models into production and owning them afterwardsComfort working on open-ended problems with no agreed definition of successNice to haveKnowledge of energy markets or energy systemsQuant trading experience, such as arbitrage, backtesting or riskHands-on work with transformer architectures, generative models or agentic AIA PhD in a quantitative subjectCompensationThe base range is £120k to £150k. Everyone in the business holds equity, and you’ll be shown a model of the projected upside during the process. The role is hybrid, with 2 days a week in the London office.Interview ProcessThe interview process has three stages: a conversation with the VP of Data Production, a technical take-home, and a final discussion with the CEO.