Data Science Lead

National e-Governance Division (NeGD), Digital India Corporation; Delhi, New Delhi, Dadri; Sur place; INR 3 000 000 - 5 200 000; Plein temps

Résumé du poste

National e-Governance Division (NeGD), Digital India Corporation, is hiring two Data Science Lead positions. You will lead a team in one specialisation, own the predictive and risk models used in government operations, and provide technical oversight of vendors.

Strong statistical depth and ability to explain results to decision makers are essential. The role requires 8+ years in data science with leadership experience, plus a track record of production-grade models and governance practices.

Qualifications

• 8+ years in data science, statistics or applied machine learning; 12+ years preferred; minimum 4 years building models in the chosen specialisation used in operational decisions
• At least 3 years leading data science teams or projects, including review of other data scientists' models
• Demonstrable ownership of at least two predictive, risk or fraud systems in production, with validation and outcomes
• Experience with causal inference or programme evaluation is a strong plus; public sector, health, agriculture, banking, insurance or payments experience is a plus

Responsabilités

• Translate business questions into well-defined problems with a clear decision, baseline and evaluation metric
• Assess data feasibility, including coverage, reporting delays, label quality, leakage and bias, and advise where data is insufficient for the objective
• Predictive Analytics: time-series, hierarchical and spatial forecasts with reconciling across levels; risk and propensity scoring; early-warning thresholds
• Fraud & Risk Analytics: define fraud typologies with investigators; use unsupervised, semi-supervised and positive-unlabelled learning
• Validate models with leakage-free schemes and quantify uncertainty via prediction intervals, calibration and conformal prediction
• Monitor drift after go-live and assess improvements via experiments or quasi-experimental methods
• Audit fairness, provide explanations for decisions, and maintain model cards and audit trails
• Deliver dashboards and analytical services on a shared metric layer with the Data Engineering Lead
• Lead in-house data scientists, review their work, and provide technical oversight of vendor teams

Connaissances

Statistics

Forecasting

Anomaly detection

Team leadership

Causal inference

Data science

Formation

B.Tech./B.E./M.Tech./M.S./M.Sc.

Ph.D. desirable

Data Science related field

Outils

PyMC

XGBoost

LightGBM

CatBoost

TabPFN

SHAP

PySpark

SQL

Description du poste

We are hiring for 2 - Data Science Lead positions. When applying, please select only ONE specialisation that best matches your expertise and experience.

Predictive Analytics

Forecasting, risk and propensity scoring, and early-warning models

Fraud & Risk Analytics

Fraud and anomaly detection, entity resolution and network analysis on claims and transactions

Role Summary

The Data Science Lead will lead a data science team in one specialisation, owning the predictive, risk and anomaly models that government departments use for operational decisions, and providing technical oversight of vendor teams. The role requires statistical depth, sound judgement on the choice of method, rigorous validation with quantified uncertainty, and the ability to explain models to senior decision makers and investigators.

Educational Qualification

• B.Tech./B.E., M.Tech./M.S. or M.Sc. in Statistics, Mathematics, Economics, Computer Science, Data Science or a related quantitative discipline
• Ph.D. or published applied work in statistics, econometrics, forecasting or fraud analytics is highly desirable
• Certifications are not mandatory; demonstrated work, as listed under Minimum Evidence at Screening, shall carry greater weight
• Candidates with strong verifiable work (Kaggle Master level or above, published research or significant open-source contributions) will be considered in lieu of formal qualification

Experience

• 8+ years in data science, statistics or applied machine learning (12+ years preferred); minimum 4 years building models in the chosen specialisation used in operational decisions
• At least 3 years leading data science teams or projects, including review of other data scientists' models
• Demonstrable ownership of at least two predictive, risk or fraud systems in production, with documented validation and measured outcomes
• Predictive Analytics: experience with hierarchical, spatial or panel data, and with missing, delayed or inconsistent data
• Fraud & Risk Analytics: experience detecting fraud or anomalies with few labels, including record linkage across sources and graph-based detection
• Experience with causal inference or programme evaluation is a strong plus; experience in the public sector, health, agriculture, banking, insurance or payments is a plus

Key Responsibilities

Problem Formulation

• Translate business questions into well-defined problems with a clear decision, baseline and evaluation metric
• Assess data feasibility, including coverage, reporting delays, label quality, leakage and bias, and advise where the data is insufficient for the objective

Specialisation Responsibilities

• Predictive Analytics: time-series, hierarchical and spatial forecasts at district, state and national level, with reconciliation across levels and backtesting; risk, propensity and prioritisation scores; early-warning thresholds designed with the operational teams who will use them
• Fraud & Risk Analytics: fraud typologies defined with investigators; rules combined with unsupervised, semi-supervised and positive-unlabelled learning; entity resolution across individuals, bank accounts, addresses and documents; graph analysis of linked cases; risk scores with reason codes and review queues sized to investigator capacity; checks against identity, tax, document and bank-account records

Validation, Uncertainty & Monitoring

• Validate models with leakage-free schemes, and quantify uncertainty through prediction intervals, calibration and conformal prediction
• Monitor drift after go-live, including adaptation by fraudsters, and assess whether deployed models improved outcomes using experiments or quasi-experimental methods

Responsible AI & Governance

• Audit fairness across regions and population groups, provide explanations for decisions affecting individuals, and maintain model cards, evaluation records and audit trails
• Apply anonymisation and data minimisation in compliance with DPDPA 2023

Analytics Products

• Deliver dashboards and analytical services on a shared metric layer, with the Data Engineering Lead and the engineering team

Team Leadership & Vendor Oversight

• Lead in-house data scientists, review their work, and remain hands‑on
• Provide technical oversight of vendor teams, and recommend acceptance or rejection of deliverables

Technical Competencies

• Statistics & Causal Inference: generalised linear, hierarchical and Bayesian models (PyMC); A/B testing, difference-in-differences, synthetic control, regression discontinuity (DoWhy, EconML)
• Machine Learning: XGBoost, LightGBM and CatBoost; tabular foundation models (TabPFN); explainability (SHAP); calibration
• Uncertainty: prediction intervals, conformal prediction (MAPIE), bootstrap methods
• Predictive Analytics: ARIMA, ETS and state-space models; hierarchical reconciliation (MinT); statsforecast, Darts or GluonTS; time-series foundation models (Chronos-2, TimesFM); GeoPandas and H3
• Fraud & Risk Analytics: PyOD, Isolation Forest, autoencoders; Splink, Zingg or Dedupe with transliteration-aware matching of Indian names and addresses; Neo4j, NetworkX, PyTorch Geometric or DGL; weak supervision and Cleanlab; precision at k and cost-sensitive metrics
• Data & Engineering: advanced SQL, PySpark, Polars or DuckDB; Python with pytest, Git, MLflow and DVC
• Communication: Superset, Power BI or Tableau; explaining limitations and uncertainty to non-technical audiences
• Governance: Fairlearn; DPDPA 2023; MeitY Responsible AI advisory