Senior Data Scientist

Senior Data ScientistLocation: Manchester (Hybrid working)Who we are:Founded in 2003, Mobysoft provides data-based insight solutions to a wide range of social housing clients, through market-leading products, simultaneously helping keep tenants housed in homes they can enjoy and improving social housing landlords long term organisational health.Our vision is a world in which intelligent technology significantly improves the quality of life for people who live in social housing and our mission is delivering accurate actionable data insights that help social housing providers ensure a consistent, equitable service.What are we looking for?We are an ambitious, customer-centric, hybrid-working Data & Analytics team, dedicated to developing a new generation of data products that unlock significant value for the social housing sector. We operate with a focused product lens, driven by curiosity and a commitment to technical excellence.As a Senior Data Scientist you will take deep, structured/tabular problems – rent arrears, tenant risk, contact strategy, repairs history – and work them through to clear, evidence-based outcomes.Alongside our existing senior data scientist (who leads on natural language processing [NLP] / and neural networks [NN]) you will be part of a small, central function that partners closely with product and engineering teams on a single, shared roadmap, with a one-team mentality throughout.Key ResponsibilitiesWhat will you be doing?Our direction of travel includes:Early-warning and lifecycle modelling. For rent, distinguishing genuine tenant arrears from technical or timing artefacts, and tracking cases from early warning signs through to resolution, stabilisation, or support; for properties, detecting risks, and understanding the balance between planned and responsive repairs.Forecasting. Forecasting rent payment patterns, balances, and repairs/asset needs over time – for example which property cohorts to prioritise for planned repairs – using both classical and modern time-series methods.Contact strategy. Designing contact strategies that focus limited capacity on where it can make the most difference across channels including human, AI-assisted, and SMS.Prescriptive analytics. Linking model outputs to prescriptive, next-best-action recommendations. For example, root cause analysis to help get first-time fixes right on repairs.AI agents. Designing, building, and deploying AI agents that combine our own data with external reference sources to support faster, better decisions.Deployment & monitoring. Deploying and monitoring machine learning (ML) models in production, with the engineering discipline to catch drift and issues early.Cross-team support. Occasional collaboration on NLP/NN-related work led by our existing senior data scientist, providing cover when needed.What's next. Plenty that hasn't been thought of yet. This list will evolve alongside our business, the role and your findings, insights and ideasIn short, there will be lots to keep you interested, opportunities to keep your technical and interpersonal skills developing, and a real chance to drive innovation and change.This is a professional, technical role with no line management responsibilities. There will be multiple opportunities to take technical ownership of, and lead the delivery of, specific data science workstreams.Qualifications and skillsRequiredWhat qualifications and experience are we after?A Master's or PhD in Data Science, ML, AI, or a related quantitative discipline – or equivalent demonstrated commercial experience.5+ years of serious commercial experience across complex problems, including transactional/event-level data, with multiple live solution deployments.Experience working in a test-and-learn way as part of a central data team collaborating closely with multiple product and engineering teams against a shared roadmap.A track record of committing to and delivering against time-boxed checkpoints, producing concrete, shippable outputs on a schedule.What technical skills are required?Data EngineeringWrangling and engineering data across our warehouse:Ability to source, validate, and shape data – including feature engineering and external/open data – grounded in a desire to genuinely understand the data-generating processes and domain context.Comfortable working across common data formats (e.g. CSV, JSON, Parquet) – loading, inspecting, joining, and aggregating as needed – and understanding what it means for the work when data is slowly changing versus arriving in near real time.CodingStrong Python and SQL.AI AgentsAbility to design and build AI agents that augment the data science process. For example, an agent that cross-references external reference data against our internal records to distinguish a genuine anomaly from a data artefact.Core Statistical & ML ToolkitUsing gradient boosting (e.g. XGBoost, LightGBM, CatBoost) as the default for tabular prediction.Recognising when simpler, more interpretable models, such as regularised logistic regression or well-constrained decision trees, are the better choice.Applying clustering and segmentation techniques (e.g. k-means, hierarchical or sequence-based clustering) to characterise populations.Model CraftStrong feature engineering skills – turning business context and raw structured data into meaningful, stable and explainable model features.Testing whether a model generalises across different populations or datasets, and re-validating or re-tuning it when it's applied somewhere new.Designing experiments so they're free of leakage.Able to recognise when data limitations, rather than modelling technique, are the constraint.Properly handling imbalanced data.Calibrating predicted probabilities so they can be understood by domain experts.Modelling Change Over TimeTime-to-event and state-transition approaches (e.g. discrete-time classification, roll-rate models) for lifecycle and hazard problems.Forecasting using classical statistical methods, Bayesian/state-space approaches, and modern pretrained time-series foundation models.Connecting Models to Real DecisionsFraming problems like contact-list generation as ranking/resource-allocation – focusing limited contact capacity where it's likely to make the most difference.Using causal inference and experiment design (e.g. A/B testing) to draw reliable conclusions from real-world, non-experimental data.Engineering DisciplineApplying solid software engineering practice to ML models – testing, version control, reproducible environments, and production-ready code.Delivering work in a time-boxed, iterative way.Explainability & FairnessConsidering fairness end-to-end – from data, feature and model choices through to testing the resulting model for bias across population groups.DesirableA background in another regulated or asset-heavy sector – for example financial services, insurance, utilities, or the public sector – that transfers well to this problem space.General familiarity with neural network architectures and NLP/large language model (LLM) tooling, sufficient to pick up key aspects of a teammate's deep-learning codebase with relative ease and do basic fault-finding when needed – this role is structured-data-first, but should be able to provide occasional cover on the team's NLP/NN work.Experience working with cloud infrastructure (ideally Amazon Web Services [AWS]) for data storage, training, and deployment.The PersonYou are someone who:Works effectively both independently (e.g. during remote deep work) and collaboratively within a team.Is genuinely curious about where data science and AI are heading, matched with the judgement to weigh new methods against business priorities, timescales, and problem fit.Communicates clearly, in writing and verbally, including with non-technical audiences.Works to understand the business context, with a proven ability to align with and actively support business goals, objectives and key results (OKRs).Looks to build domain knowledge within the sector of application as an intrinsic part of doing good data science.Is focused on shipping and delivering value, as part of a team that shares that discipline.MobyIdeals:You will behave in accordance with the MobyIdeals:Customer-focused: We drive outcomes that create value for the customer. We continually challenge ourselves on ‘what’s in it for the customer’. We drive win/win/win solutions.Collaborative: We operate as one, fostering open communication, diverse contribution, cooperation and trust. We inspire teams towards a common goal for success.Outcome-orientated: We are driven by the end goal, rather than the process or steps to get there.Accountable: We own decisions, are transparent, set clear expectations and consistently deliver on commitments.Courageous: We actively contribute and constructively challenge with positive intent. We think big and move at pace.Innovative: We own and proactively search for solutions. We positively embrace problems and lead change.If you are interested and would like to know more then please apply to simone.ryan@mobysoft.com or via our careers page. Please note that we are not working with any external Agencies for this position.