Senior Machine Learning Engineer - New Verticals Agentic Foundations
About this role
Building scalable AI agents that handle complex grocery and retail tasks requires closing the loop between real-world usage and foundational model improvements. The Agentic Foundations team at DoorDash creates the core intelligence powering assistants like the Ask grocery tool, and this role directly addresses the challenge of making those agents faster, cheaper, and more reliable. Day-to-day work centers on refining evaluation loops, optimizing prompts and tools, and balancing quality, latency, and cost through rigorous experimentation.
Engineers joining this group will collaborate closely with peers focusing on agentic memory, product representations, and small language model post-training. Success in the position requires a strong background in NLP, practical experience building LLM-based agents, and a collaborative mindset geared toward shipping production-ready systems that directly impact millions of consumers, merchants, and dashers.
What you will do
- Own the continuous improvement of flagship consumer agents by utilizing an evaluation-driven optimization loop to enhance speed, cost-efficiency, and reliability.
- Conduct rigorous experiments involving the fine-tuning of open-weights models and the deployment of small language models to replace or augment larger alternatives.
- Take new agentic opportunities for merchants and dashers from early exploration stages all the way to production deployment.
- Partner closely with teammates working on agentic memory, product representations, and post-training to feed production insights back into foundational models.
- Collaborate with engineering, product, and business leaders to define the machine learning strategy for the grocery and retail delivery sector.
What you need
- Bring at least 3 years of industry machine learning experience with hands-on practice building and shipping LLM-based agents involving tool use, context management, prompting, and guardrails.
- Demonstrate experience with post-training or fine-tuning of open-weights models, such as supervised fine-tuning, preference optimization, or reinforcement learning.
- Maintain a strong foundation in natural language processing and machine learning, coupled with proficiency in Python and frameworks like PyTorch or TensorFlow.
- Hold an M.S. or PhD in Computer Science, Statistics, Math, or another quantitative field, or possess equivalent practical experience.
Benefits
- 401(k) plan with employer matching
- 16 weeks of paid parental leave
- Wellness benefits and commuter benefits match
- Flexible paid time off and paid sick leave
- Medical, dental, and vision insurance
- Disability, basic life insurance, and a mental health program
Skills and tech stack
Questions candidates ask
What is the remote work policy for this position?
The role is expected to be hybrid with some time spent working in-office and some time remote.
What is the compensation range for the role?
The starting base pay falls within a range of $137,100 to $201,600 USD, with exact figures localized according to work location and other job-related factors. Additionally, compensation includes opportunities for equity grants.
What kind of technologies and models will I be working with?
Work involves large language models, small language models, post-training and fine-tuning of open-weights models, Python, and frameworks such as PyTorch or TensorFlow.
What team will I be joining?
You will join the Agentic Foundations team, which is part of New Verticals ML, and report directly to the engineering lead on the New Verticals AI/ML team.