Software Engineer, ML Infrastructure (MDX)

DoorDash USA•Sunnyvale, CA

15h•$130,600 - $192,000

About The Position

DoorDash's MDX team, within Machine Learning Platform, builds the foundational infrastructure for ML engineers across DoorDash, Wolt, and Deliveroo. They cover the full ML lifecycle, from feature engineering to monitoring, with a mission to enhance the speed and reliability of ML-driven business impact. A key focus is agentic infrastructure, developing systems that allow AI agents to accelerate and manage ML workflows, including auto research for exploring modeling ideas and self-healing systems for detecting and resolving failures in ML pipelines and production models with minimal human intervention. This role involves joining a small, high-leverage team to build the agentic layer of DoorDash's ML infrastructure. The engineer will design and build systems enabling agents to speed up how ML engineers develop, ship, and operate models. This includes auto research systems for proposing, running, and evaluating modeling experiments, and agentic developer support for tasks like setting up training jobs, debugging pipelines, reviewing deployments, and navigating the platform. The role requires owning components end-to-end, from design to production, covering agent orchestration, tooling, ML platform integrations, agent behavior evaluation, and observability. It's ideal for an engineer with a strong backend foundation, enthusiasm for applying AI agents to engineering workflows, quick learning ability, and a desire for technical leadership in a rapidly evolving field.

Requirements

B.S., M.S., or PhD in Computer Science or equivalent
2+ years of industry experience in software engineering
Hands-on experience building ML infrastructure in production including feature engineering and feature stores, data pipelines for ML, training and experimentation platforms, model evaluation, deployment, serving and inference, or monitoring
Strong backend engineering fundamentals, especially in Python and distributed systems
Experience operating systems in production, including testing, observability, debugging, and reliability
A good understanding of how ML engineers and data scientists work, and a track record of building platforms, tools, or services that make them more productive
Ability to take ambiguous, well-scoped problems from idea to production, and to learn quickly in a fast-moving technical area
Proficiency in using AI coding tools (e.g., Claude Code, Codex, Cursor) across the software development lifecycle, including designing, generating code, testing, monitoring, and releasing software

Nice To Haves

Experience building LLM-powered applications or AI agents, including tool use, orchestration, or MCP servers, in production or on substantial projects
Experience with agent evaluation, LLM observability, tracing, or building eval harnesses
Experience with ML platforms or MLOps tooling (e.g., Kubeflow, MLflow, Ray, Airflow)
Experience with Kubernetes and cloud infrastructure (AWS/GCP)
Experience building developer tools, internal platforms, or self-serve infrastructure
Experience with guardrails, safety, or cost controls for LLM systems

Responsibilities

Build agentic systems that increase the velocity of ML development across DoorDash, Wolt, and Deliveroo, helping ML engineers get from idea to production model faster.
Design and implement infrastructure that lets agents propose, run, and evaluate modeling experiments.
Build agentic developer support for the ML lifecycle, such as agents that help engineers set up training jobs, debug failing pipelines, review model deployments, and navigate our platform.
Develop evaluation and observability for agent behavior, so we can measure whether agents are actually helping, catch regressions early, and build the trust needed to give them more autonomy.
Work with production standards in mind, including reliability, safety guardrails, cost controls, monitoring, and clear playbooks, so that agentic workflows can run against real ML systems.
Partner closely with ML engineers, data scientists, and platform teams to understand where their time goes and turn those workflows into durable, reusable platform capabilities.
Grow quickly in a fast-moving area, with the chance to shape how agents are used to build and operate ML systems at a company of this scale.