Staff Machine Learning Engineer, AI Platform & Agentic Apps
About this role
Designing and building the core agent harness at Robinhood involves establishing orchestration, tool integrations, and context management to safely power both internal and customer-facing agents. Operating within the AI Platform & Agentic Apps team, the person in this position will architect trajectory-level evaluation systems and action guardrails such as permission models, approval gates, and sandboxing. The group focuses on applying frontier technologies to major financial challenges, creating systems where agents take real-time actions instead of simply answering questions.
Robinhood maintains a high-performing and fast-moving environment where expectations and rewards are both significant. Candidates who excel here possess deep technical expertise in machine learning and distributed systems, alongside a strong track record of shipping agentic applications to production at scale. The role functions as a technical anchor, requiring collaboration across product, infrastructure, and engineering teams to define technical direction and mentor peers.
What you will do
- Design and build the core of the agent harness covering orchestration, tool integrations, and memory management.
- Ship agentic applications end to end from ambiguous problems to production agents that execute real actions.
- Build trajectory-level evaluation systems that score how agents reach answers through multi-step task completion and planning.
- Architect action guardrails as platform primitives including least-privilege tool scoping and sandboxing.
- Turn evaluations and guardrails into adoptable products for other teams via SDKs and CI regression gates.
- Set the technical bar through architecture reviews, code reviews, and mentorship while defending technical decisions with data.
What you need
- Bring 10+ years of experience as a machine learning engineer or ML-focused software engineer with strong Python and distributed-systems fundamentals.
- Hold a Master's degree in Computer Science or a related technical field, or demonstrate equivalent professional experience.
- Possess hands-on experience building agentic systems end to end on top of frontier models in production.
- Demonstrate deep expertise evaluating agents by building trajectory-level evals, tool-call scoring, and simulation environments.
- Show expertise in designing action-level guardrails like permission models and blast-radius controls for agents in sensitive systems.
- Apply rigor in evaluation methodology including golden datasets, rubric grading, statistical significance, and contamination control.
- Prove the ability to build platforms and tooling that other engineering teams successfully adopt.
Benefits
- Performance driven compensation with multipliers, bonus programs, equity ownership, and 401(k) matching
- 100% paid health insurance for employees with 90% coverage for dependents
- Access to the Robinhood Employee Fund for eligible US employees
- Access to the best AI tools on the market and continuous AI skill-building
- Lifestyle wallet for wellness, learning, and flexible spending
- Employer-paid life and disability insurance, fertility benefits, and mental health benefits
Skills and tech stack
Questions candidates ask
What is the remote work policy for this role?
This role is based in the Menlo Park, CA office, with in-person attendance expected at least 3 days per week. Robinhood believes in the power of in-person work to accelerate progress, spark innovation, and strengthen community.
What is the salary range for this position?
The expected base pay range for this role in Menlo Park, CA is $255,000 to $300,000 USD. Base pay depends on job-related factors including education, training, experience, location, business needs, or market demands, and the role is also eligible for bonus opportunities, equity, and benefits.
What will I actually work on day to day?
You will design and build the core of Robinhood's agent harness, including orchestration, tool integrations, and context management. You will also build trajectory-level evaluation systems and architect action guardrails as platform primitives.
What kind of background does the team look for?
The team looks for 10+ years of experience as a Machine Learning Engineer or ML-focused software engineer with strong Python and distributed-systems fundamentals. Candidates should have a track record of shipping LLM-powered systems and agentic applications to production at scale.