> Markdown version of [/jobs/ext/2579987-member-of-technical-staff-lead-researcher](https://www.wearedevelopers.com/jobs/ext/2579987-member-of-technical-staff-lead-researcher). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Member of Technical Staff, Lead Researcher - **Company:** DOORDASH, INC. - **Location:** San Francisco, CA, United States - **Experience:** Expert - **Salary:** $203,500.0 - $299,300.0 - **Contract:** Permanent contract - **Skills:** Artificial Intelligence, Learning Management Systems, Open Source Technology, Azure Machine Learning, Video Capture, Large Language Models, Multi-Agent Systems, Operational Systems, Machine Learning Operations - **Published:** August 3, 2026 - **Apply:** https://www.indeed.com/viewjob?jk=08eb231ea369cd6b ## About the Role * A strong research track record: first-author publications at top ML venues, or equivalent demonstrated output (widely-used systems, influential open-source work, or research artifacts adopted at scale) * Experience leading ambitious research projects end-to-end, from problem framing through to results that other people built on * Taste: the ability to identify which problems are worth working on, when an approach is exhausted, and when a result is real * A builder's instinct: comfortable working close to real systems and real data, not just on benchmarks * Excitement about the founding-team aspects of the role: hiring, agenda-setting, culture-building, and partnering across functions * Impact, you want to ship frontier systems to production and see your work being leveraged by hundreds of millions of consumers * Excellent written and verbal communication, and a track record of mentoring or developing other researchers PhD in a relevant field is typical but not required. We weigh demonstrated research output and judgment far more heavily than credentials. ## Description * Lead high-impact research projects end-to-end - from problem framing through publication and, where appropriate, production deployment * Help build the team - interview, recruit, and mentor researchers, engineers, and fellows joining the org * Shape the org's culture and operating model - how we publish, how we collaborate with product teams, how we balance open research with proprietary work * Partner across DoorDash with ML platform, product, and operations teams to identify the highest-leverage research bets and translate findings into real-world impact What You'll Have Access To * Novel proprietary data at marketplace scale - logistics traces, merchant operations, consumer behavior, real-time supply and demand signals, and longitudinal data unavailable anywhere else * Scalable data collection - ability to design and run structured data collection, leveraging DoorDash's world-class operational scale, from in-the-wild image and video capture to operational task demonstrations and human-in-the-loop annotation, at a scale and physical-world coverage no other org can match * High compute budgets for training and inference, sized to support frontier-scale experimentation including large-model pre-training and post-training, RL training runs, and large-scale evaluation sweeps * Full research infrastructure - DoorDash's internal RL stack, RL environments built on real operational systems, training and evaluation pipelines, and agent evaluation harnesses, with engineering support to extend them as your research demands * Direct access to leadership - a seat at the table for the decisions that shape the research org, with the autonomy to operate as a principal-level researcher * Publication freedom - we expect and support publication at top venues (NeurIPS, ICML, ICLR, RSS, CoRL, KDD, etc.) with a fast, supportive internal review process * Compute and data for external collaborators - budget to bring in academic collaborators, fellows, and visiting researchers as your agenda requires Research Areas We are broadly interested in researchers across the following areas, though the right candidate may reshape this list: * Agentic systems for logistics and local commerce - long-horizon planning, tool use, multi-agent coordination, and evaluation methodologies for agents operating in physical-world marketplaces * Memory and personalization - transfer RL, continual learning, harness-based improvements, and systems that adapt to individual consumers, merchants, and Dashers over time without catastrophic forgetting or unsafe drift * Foundation models for marketplace dynamics - forecasting, pricing, matching, and personalization at marketplace scale, including domain-specific pre-training and post-training * Evaluation and measurement - new benchmarks, eval harnesses, and methodologies for ML systems deployed in messy, real-world operational settings * Multimodal understanding - vision, speech, and language applied to merchant catalogs, in-store and on-the-road imagery, and consumer interfaces * Robotics and embodied AI for last-mile delivery - perception, planning, and learning systems for the physical edge of the marketplace ## Related Videos - [The Data Phoenix: The future of the Internet and the Open Web](https://www.wearedevelopers.com/videos/1116-the-data-phoenix-the-future-of-the-internet-and-the-open-web) - [The Fundamentals of Online Video](https://www.wearedevelopers.com/videos/32-the-fundamentals-of-online-video) - [Enterprise Linux as Container Images](https://www.wearedevelopers.com/videos/1610-enterprise-linux-as-container-images) - [Designing and Deploying Distributed Multimodal Multi-Agent Systems with Google's AI Stac](https://www.wearedevelopers.com/videos/1976-designing-and-deploying-distributed-multimodal-multi-agent-systems-with-google-s-ai-stac) - [You are not an AI developer](https://www.wearedevelopers.com/videos/1148-you-are-not-an-ai-developer) - [Edge Orchestration for the Physical World: Connecting Cameras, Sensors, and Devices with MQTT](https://www.wearedevelopers.com/videos/100081-edge-orchestration-for-the-physical-world-connecting-cameras-sensors-and-devices-with-mqtt) ## Related Articles - [Navigating the AI Shift](https://www.wearedevelopers.com/magazine/629-navigating-the-ai-shift) - [How to Become an AI Engineer](https://www.wearedevelopers.com/magazine/331-how-to-become-an-ai-engineer) - [Coffee with Developers - Maria Apazoglou - Making AI understandable for all in production](https://www.wearedevelopers.com/magazine/475-coffee-with-developers-maria-apazoglou-making-ai-understandable-for-all-in-production) - [Dev Digest 137 - AI'm not sure about this](https://www.wearedevelopers.com/magazine/485-dev-digest-137-ai-m-not-sure-about-this) - [Never delegate the understanding](https://www.wearedevelopers.com/magazine/749-never-delegate-the-understanding) - [Résumé-Driven Development: How IT trends affect the job market for software developers](https://www.wearedevelopers.com/magazine/59-resume-driven-development-how-it-trends-affect-the-job-market-for-software-developers)