Principal Data Scientist builds and scales causal-inference models for promotion optimization, uplift modeling, and budget-constrained allocation in a large Southeast-Asian marketplace.
Own the end-to-end modelling and estimation behind promotion optimisation — elasticity, heterogeneous treatment effects, budget-constrained allocation, and increamentality — from problem framing to production.
Advance our causal inference stack: experiment and quasi-experiment design (geo/switchback tests, holdouts, diff-in-diff, synthetic control), debiasing observational data, and variance reduction — raising the bar on experimentation rigor across the team.
Design and productionize heterogeneous treatment effect (uplift) models at scale (tens of millions of users), with honest offline evaluation (uplift/Qini curves, policy-value estimation).
Formulate and solve budget-constrained allocation under fairness and dynamic business constraints — from LP/MILP to greedy or Lagrangian methods where they scale better.
Mentor and technically guide a team of data scientists; set standards for model evaluation, documentation, and scientific review.
Partner with business to turn model outputs into budget decisions, and communicate tradeoffs (subsidy efficiency vs. growth) clearly.
What You Will Need
8+ years in data science or ML, with 3+ years focused on causal inference or uplift modelling in production settings.
Deep expertise in heterogeneous treatment effect estimation, with hands-on production experience in several of: meta-learners (S/T/X/R), causal forests, DR-learner, or deep uplift architectures.
Hands-on experience optimising promotions, pricing, or marketing incentives with evolving constraints and measurable business outcomes.
Strong grounding in experimentation and observational causal methods — propensity weighting, instrumental variables, synthetic control, difference-in-differences.
Experience with constrained optimisation (MILP, Lagrangian methods, etc.) applied to resource allocation.
Proficiency in Python and SQL; shipping models to production with engineering partners.
Track record of technical leadership at principal/staff level: setting technical direction for a team, reviewing high-stakes analyses, and influencing roadmaps and partner teams without direct authority.
Experience in marketplaces, ride-hailing, food delivery, e-commerce, or fintech.