AI Research Fellowship

About the Team

The DoorDash Research Fellowship is a 3-month program (extendable to 6 months) looking for Summer and Fall 2026 cohorts, for researchers and engineers who want to work on the hardest applied ML and AI problems in local commerce. Fellows are given the resources, autonomy, and access to real-world operational data needed to pursue ambitious research directions — with the goal of producing work that influences both the field and how DoorDash operates at scale.

This program is modeled on the best external research fellowships: fellows are treated as independent researchers, not as junior employees on a product team. You pick the problem (within a set of priority areas), you own the direction, and you publish or ship the outcome.

You’re excited about this opportunity because you will receive…

• Dedicated compute allocation sized to the research agenda — GPU clusters for training and inference budgets for experimentation
• Full access to DoorDash's research infrastructure — our internal RL stack, training and evaluation pipelines, RL environments built on real operational systems, agent evaluation harnesses, and the tooling our own research teams use day-to-day. Fellows are first-class users, not sandboxed visitors.
• Access to DoorDash operational data — real-world datasets spanning logistics, merchant operations, consumer behavior, and marketplace dynamics, under appropriate data governance
• Research mentorship from senior researchers and engineering leaders at DoorDash, plus a named research sponsor for each fellow who meets with you weekly and is accountable for unblocking your work
• Speaker series featuring leading researchers and practitioners from academia and industry — faculty from top ML programs, research leads from frontier AI labs, and senior operators from across tech. Fellows get dedicated 1:1 time with speakers when possible.
• A cohort of fellows working alongside you — a small, tight-knit group of researchers tackling different problems but sharing ideas, drafts, and reading groups. The community extends to DoorDash research alumni and the broader fellowship network.
• Publication support — fellows are encouraged to publish at top venues (NeurIPS, ICML, ICLR, KDD, etc.) and DoorDash will support the legal and review process
• Relocation support, housing stipend, and competitive pay

Structure

• Kickoff (2 weeks, in-person in SF): Onboarding, scoping the research agenda with your sponsor, meeting teams across Research, ML Platform, and relevant product orgs
• Core research period (~10 weeks, hybrid from SF DoorDash office): Focused research work with regular check-ins, internal talks, and working sessions. Extension to 6 months decided at the midpoint based on research progress and mutual fit.
• Closeout: Final write-up, internal presentation, and publication or productionization path

Priority Research Areas

Fellows are welcome to propose their own direction, but we are particularly interested in work across:

• Reinforcement learning environments for real-world operations — building high-fidelity simulators and training environments from operational data, including the tradeoffs between real-data fidelity and synthetic generalization
• Agentic systems for logistics and local commerce — long-horizon planning, tool use, and evaluation methodologies for agents operating in physical-world marketplaces
• Foundation models for marketplace dynamics — forecasting, pricing, matching, and personalization at marketplace scale
• Evaluation and measurement — new benchmarks and evaluation methods for ML systems deployed in messy, real-world operational settings
• Multimodal understanding — vision, speech, and language applied to merchant catalogs, operations, and consumer interfaces

We’re excited about you because…

• You’re a researcher with a strong track record — PhD candidates (rising 4th year or beyond), recent PhDs, or independent researchers with published work at top ML venues or equivalent demonstrated output
• You’re an engineer with deep research instincts who have shipped ambitious ML or systems work and want dedicated time to pursue a research direction
• You can operate independently — set a research agenda, scope it, and execute without heavy supervision
• You have strong written and verbal communication; comfort presenting work to mixed research and engineering audiences

We are not looking for specific years of experience or specific degrees. Exceptional candidates from non-traditional backgrounds are strongly encouraged to apply.