Staff Machine Learning Scientist, Applied Causal Inference

DoorDash USA•Los Angeles, CA

2d•$203,500 - $299,300

About The Position

DoorDash is building the next generation of causal decisioning systems for New Verticals: grocery, convenience, retail, alcohol, pets, flowers, and other emerging categories. These businesses operate in high-dimensional, messy marketplaces where every consumer, merchant, item, promotion, substitution, search result, and delivery promise creates a causal question. We are hiring a Causal Machine Learning Engineer to help build the causal ML foundation behind how DoorDash grows New Verticals. This is not a generic ML role with some experimentation work on the side. We are looking for someone who has built or deeply worked on production causal systems: uplift models, heterogeneous treatment effect models, surrogate metrics, experimentation platforms, counterfactual policy evaluation, promotion optimization, or marketplace decisioning systems. You will join a small, senior pod of causal ML and econometrics experts working across ML, Analytics, Product, and Engineering. The mandate is to build the causal spine for a large-scale consumer marketplace.

Requirements

Deep practical experience with causal inference, econometrics, experimentation, or causal ML.
Experience shipping models or decision systems in production, ideally in consumer marketplaces, ads, recommendations, search, pricing, promotions, logistics, fintech, or other high-scale settings.
Strong judgment around the tradeoffs between randomized experiments, observational estimation, and model-based decisioning.
Comfort debating and applying methods such as doubly robust estimation, double ML, IV, diff-in-diff, CUPED, uplift modeling, contextual bandits, and off-policy evaluation.
Strong ML engineering ability: you can build reliable pipelines, train models, evaluate them rigorously, and partner with platform teams to put them into production.
Strong product judgment: you can connect methods to business decisions, not just optimize offline metrics.
The ability to operate across functions with ML engineers, economists, data scientists, product managers, and business leaders.

Responsibilities

Design, build, and productionize causal ML systems that influence real marketplace decisions across New Verticals.
Build uplift / heterogeneous treatment effect models for consumer lifecycle value, promotions, retention, and reactivation.
Develop counterfactual evaluation frameworks for ranking, recommendations, search, promotions, substitutions, and marketplace interventions.
Build systems that connect experimentation, observational data, and ML decisioning so teams can make better tradeoffs when randomized experiments are slow, noisy, or incomplete.
Design surrogate metrics and early indicators that help teams move faster while preserving long-term marketplace health.
Partner with econometrics and analytics leaders to choose the right methods: doubly robust estimation, IV, diff-in-diff, synthetic controls, double ML, CUPED-style variance reduction, contextual bandits, off-policy evaluation, and related approaches.
Translate causal models into production systems that can shape decisions in ranking, targeting, budget allocation, inventory-aware discovery, and consumer growth.
Raise the bar for causal reasoning across ML teams: when to trust a model, when not to, and how to debug causal claims in a real marketplace.