Senior ML Engineer: Agentic Foundations & LLMs

DoorDash USA; San Francisco, Sunnyvale, Seattle, New York (CA, CA, WA, NY); Hybrid; USD 137,000 - 202,000; Full time

Benefits offered by this job

Equity grants

401(k) with employer matching

Paid parental leave

Wellness benefits

Commuter benefits

Paid time off

Paid sick leave

Medical, dental, and vision

Job summary

DoorDash seeks a Senior Machine Learning Engineer for the Agentic Foundations team in a hybrid role. You will own continuous improvement of flagship agents, starting with the Ask Assistant, using eval-driven loops to speed, cost-efficiency, and reliability for millions of customers.

You will extend these capabilities to merchants and dashers, turning ideas into production‑ready agents. You will work at the intersection of LLMs, agents, post‑training, and evals, collaborating on agent memory,

Qualifications

• 3+ years of ML experience, incl. shipping LLM-based agents and evals
• Post-training or fine-tuning of open-weight models, including small LLMs
• Strong NLP/ML foundation; Python + PyTorch or TensorFlow

Responsibilities

• Improve flagship agents via eval-driven loop; optimize prompts, tools, skills, and context
• Balance quality, latency, and cost; fine-tune models where needed
• Launch new agentic experiences for Merchants and Dashers
• Bridge foundations to production agents with memory and product representations
• Shape ML-driven roadmap for grocery and retail delivery growth

Skills

3+ years ML experience

LLM-based agents

Tool use

Context management

Prompting

Guardrails

Eval-driven improvements

Education

MS/PhD in CS/Stats/Math or equivalent

Tools

Python

PyTorch

TensorFlow

Job description

DoorDash seeks a Senior Machine Learning Engineer for the Agentic Foundations team in a hybrid role. You will own continuous improvement of flagship agents, starting with the Ask Assistant, using eval-driven loops to speed, cost-efficiency, and reliability for millions of customers.

You will extend these capabilities to merchants and dashers, turning ideas into production‑ready agents. You will work at the intersection of LLMs, agents, post‑training, and evals, collaborating on agent memory,