Senior Machine Learning Engineer, New Verticals Agentic Foundations

The posting, as DoorDash wrote it

About the Team

Agentic Foundations, part of New Verticals ML, builds the core AI/ML intelligence that powers high-quality, scalable agents at DoorDash, and proves it through flagship agent experiences such as the Ask grocery assistant. We operate as a closed loop: new models, evals, and representations make our agents better, and real-world agent usage surfaces the gaps that drive our next foundations work.

About the Role

We’re looking for a Senior Machine Learning Engineer to join Agentic Foundations with a primary focus on Vertical Agent Development. You will own the continuous improvement of our flagship agents, starting with the Ask Assistant, using an eval-driven loop to make them faster, more cost-efficient, and more reliable for millions of customers. You will also help us bring this approach to new agentic experiences for Merchants and Dashers, turning promising ideas into production-ready agents.

This is a chance to work at the intersection of LLMs, agents, post-training, and evals on problems that ship. You will partner closely with engineers working on agentic memory, agent-compatible product representations, and post-training of small language models (SLMs), so that what we learn from production agents feeds directly back into our foundations, and what we build there makes our agents better.

You will report into the engineering lead on our New Verticals AI/ML team. We expect this role to be hybrid with some time in-office and some time remote.

You’re excited about this opportunity because you will…

• Raise the quality of our Consumer agents: build and refine our eval harness-optimization loop to pinpoint failure modes, then iterate on prompts, tools, skills, and context, and measure the impact on real customer tasks.
• Find the best balance of quality, latency, and cost: run rigorous experiments, including fine-tuning open-weights models and deploying small language models (SLMs) where they can replace or augment larger LLMs.
• Launch new agents: take agentic opportunities for Merchants and Dashers from early exploration to production.
• Turn foundations into product impact: work with teammates on agentic memory, agent-compatible product representations, and post-training SLMs for steerable generative recommendation, and bring those capabilities into production agents.
• Shape the roadmap: partner with engineering, product, and business leaders to define an ML-driven strategy for our fast-growing grocery and retail delivery business.

We’re excited about you because you have…

• 3+ years of industry ML experience, including hands-on work building and shipping LLM-based agents (tool use, context management, prompting, guardrails) and improving them with evals and data
• Experience with post-training or fine-tuning of open-weights models (e.g., SFT, preference optimization, or RL), ideally including small language models, and sound judgment on quality, latency, and cost trade-offs
• Strong foundation in NLP and machine

[…]
success have no place at DoorDash. We value a diverse workforce – people who identify as women, non-binary or gender non-conforming, LGBTQIA+, American Indian or Native Alaskan, Black or African American, Hispanic or Latinx, Native Hawaiian or Other Pacific Islander, differently-abled, caretakers and parents, and veterans are strongly encouraged to apply. Thank you to the Level Playing Field Institute for this statement of non-discrimination.

Pursuant to the San Francisco Fair Chance Ordinance, Los Angeles Fair Chance Initiative for Hiring Ordinance, and any other state or local hiring regulations, we will consider for employment any qualified applicant, including those with arrest and conviction records, in a manner consistent with the applicable regulation.

If you need any accommodations, please inform your recruiting contact upon initial connection.

Notice Regarding Use of AI and Automated Tools: To streamline our hiring process, DoorDash utilizes an automated recruitment tool called Metaview.

How it works: Metaview assists our recruiting team by evaluating job related qualifications and characteristics in connection with hiring. The tool is designed and used to support - rather than replace - human decision-making; trained personnel make final decisions with meaningful human review and oversight, and DoorDash does not use Metaview or other AI-enabled tool in a manner that has the effect of subjecting applicants or employees to discrimination based on any protected characteristic or proxy or for engaging in any protected activity under applicable law.

Data Retention, Privacy & Bias Audit: Data collected during this process is retained in accordance with our Candidate Privacy Policy https://help.doordash.com/en-us/legal/article/ax-privacy-notice and applicable state laws. In compliance with New York City Local Law 144, the independent bias audit summary for Metaview is publicly available for review at https://trust.warden-ai.com/metaview/application-review .