Credit & Risk Strategist (B2C, Fintech)

We are looking for a rigorous, systems-minded quantitative leader to own our core commercial intelligence engine end-to-end. This is not a traditional dashboard-building or BI reporting seat. You will define, build, and encode the mathematical logic, unit-economic models, and predictive forecasting systems that power our product and drive real commercial decisions for our brands.

You will sit at the intersection of finance, data science, and product engineering, turning complex commercial data, messy founder inputs, and predictive models into defensible, automated product logic and AI agent workflows.

Key Accountabilities

1. Proprietary Systems: Growth Forecasts & P&Ls

• Growth Forecasting: Build and own the methodology that takes a brand’s current performance and projects a defensible 12-month forecast, complete with a rigorous assumption set.
• Financial Architecture: Construct brand-level P&Ls directly from the forecast, mapping revenue, COGS, contribution margin, and operating costs.
• Variance & Calibration: Continuously run models against real outcomes, measure errors, track actuals against projections, and explain every variance.

2. Productization & Agentic Workflow Design

• Product Logic: Translate sophisticated financial and statistical models into production-ready product logic so the system calculates and delivers outputs natively.
• AI Agent Specification: Define what data agents compute, what internal data/benchmarks they draw on, and what assertions they are strictly barred from making.
• Human-in-the-Loop Orchestration: Replace manual validation with automated product rules while establishing clear review gates for human sign-off before deliverables ship.
• Data Schemas & Engineering Collaboration: Specify data structures, validation rules, and pipelines across structured commercial inputs, partnering directly with engineering to bring models to life in production.

3. Decision Framing & Commercial Rigor

• Decision-First Modeling: Design every analysis backwards from the specific commercial decision it is meant to change.
• Uncertainty & Stated Confidence: Size uncertainty statistically, state clearly what sparse or self-reported data cannot support, and attach a stated confidence level to every output.
• Benchmark Library Management: Define and version the benchmark repository (category, stage, market) to evaluate brands accurately even when data is sparse.

4. Enablement & Commercial Communication

• Cross-Functional Communication: Present complex numbers, models, and uncertainties to founders and operators with clarity and authority.
• Quantitative Literacy: Raise the analytical bar across the business by teaching teams how to read and interpret model outputs correctly.
• Documentation & Standards: Write airtight documentation and standards for data/finance deliverables so cross-functional delivery teams can execute autonomously.

What This Seat Is Not

• Not a BI/Reporting Seat: Dashboards are merely one downstream output of this work; your primary focus is proprietary modeling and product architecture.
• Not Research Without Action: We judge models by the decisions they actually changed, not academic complexity.
• Not a Data Engineering Seat: You will specify pipelines, schemas, and logic for engineering to build—you won't be maintaining core infrastructure tables yourself.

Ideal Background & Qualifications

Must-Haves

• Experience: 6–8 years in growth, marketing, commercial, or product analytics within DTC, e-commerce, marketplaces, or specialized platforms.
• Proven Track Record: End-to-end ownership of a forecasting, unit-economics, or media-mix model that commercial teams relied on for high-stakes budget or strategic decisions. You can name the decisions your model changed and what followed.
• DTC Unit Economics Mastery: Deep fluency in CAC, AOV, COGS, contribution margin, LTV, payback period, breakeven, retention cohorts, and working capital reconciliation.
• Modelling with Incomplete Data: Demonstrated ability to build robust outputs from banded, self-reported, or partial founder inputs while explicitly quantifying confidence levels.
• Statistical Rigor: Strong foundation in sizing uncertainty, managing sample limitations, and maintaining analytical integrity.
• Technical Depth: Hands-on daily fluency in SQL and Python.
• Product & Engineering Collaboration: Experience working directly with engineering teams to turn quantitative models or metric definitions into production software.
• Communication: Exceptional presence and ability to defend underlying assumptions and model uncertainty under direct challenge.

Nice-to-Haves

• Media mix modeling (MMM), incrementality testing, or causal measurement methodologies.
• Applied machine learning experience (forecasting, propensity, or segmentation models in production).
• Experience evaluating generative AI or multi-agent system outputs.
• Familiarity with modern data stack layers (Snowflake, BigQuery, dbt) and attribution platforms (Triple Whale, Northbeam, Rockerbox).
• Exposure to multi-brand ecosystems via agency, consulting, or platform environments.
• Experience in early-stage product environments where models and products were co-developed from scratch.

Job Summary

The Data Scientist / Analyst Consultant will design, develop and deploy advanced analytics and AI/ML models to detect fraud, waste and abuse across the AB PM-JAY ecosystem. The role involves working with large volumes of claims, beneficiary, provider and transaction data, engineering fraud features, building and validating detection triggers, and converting model outputs into investigation-ready suspicious claims and cards. The Data Scientist will build triggers using advanced analytics techniques including NLP, OCR, Social Network Analysis and deep learning, support their validation with the National Anti-Fraud Unit (NAFU), and continuously improve model efficacy based on feedback and confirmed fraud outcomes.

Key Responsibilities

• Design, develop and deploy statistical and machine learning models for fraud, waste and abuse detection on scheme claims and card data.

• Perform advanced data analytics including clustering, classification, profiling, anomaly detection, predictive modelling, scoring, regression, link prediction, co-occurrence grouping, similarity matching and causal modelling.

• Develop fraud detection triggers using advanced analytics techniques such as NLP, OCR, Social Network Analysis and deep learning, and document trigger construct, logic and thresholds for review and approval.

• Apply both supervised and unsupervised learning approaches, training models on labelled historical fraud data and scheme ecosystem data.

• Conduct data profiling, quality assessment, cleaning, standardisation and feature engineering to create analysis-ready datasets.

• Execute triggers and models on claim and card data within defined turnaround timelines and generate case-level suspicious output with trigger reasons and evidence fields.

• Validate model outputs, assess false positive risk, calibrate thresholds and measure trigger efficacy against defined performance benchmarks. • Fine-tune and retrain models on the basis of subsequent events, reviewer feedback and confirmed fraud outcomes.

• Collaborate with medical experts, business analysts, subject matter experts and technology teams to translate domain knowledge and clinical rules into executable analytical logic.

• Prepare model documentation, trigger submission packs, dashboards and reports for internal review committees and client-facing evaluations. 

Preferred Skills

• Experience in healthcare analytics, health insurance claims analytics or fraud, waste and abuse detection.

• Experience in fraud analytics roles with a proven track record of developing and implementing fraud detection models and strategies.

• Knowledge of AB-PMJAY / health benefit packages / claims adjudication workflows.

• Exposure to Social Network Analysis, graph analytics or link analysis for collusion and network detection.

• Familiarity with OCR, image analytics and document forgery or duplication detection.

• Experience with model deployment, versioning and MLOps practices.

• Exposure to dashboarding and BI tools such as Power BI, Tableau or equivalent. 

Experience

• Minimum 5 years of experience in analytics and managing big data, out of which at least 3 years in developing AI/ML models.

• Demonstrable credentials of having worked on and developed AI/ML models independently.

• Experience in fraud analytics, anomaly detection or risk modelling will be preferred.

• Experience in healthcare, health insurance or government health scheme data will be an added advantage.