> Markdown version of [/jobs/ext/4583825-staff-machine-learning-engineer-ai-platform-agentic-apps](https://www.wearedevelopers.com/jobs/ext/4583825-staff-machine-learning-engineer-ai-platform-agentic-apps). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Staff Machine Learning Engineer, AI Platform & Agentic Apps

- **Company:** Robinhood
- **Location:** Menlo Park, CA, United States
- **Experience:** Expert
- **Salary:** $255,000.0 - $300,000.0
- **Contract:** Permanent contract
- **Skills:** Artificial Intelligence, Code Review, Memory Management, Python (Programming Language), Machine Learning, Management of Software Versions, Large Language Models, Agentic-AI, Build Management, AI Platforms, Information Technology
- **Published:** November 5, 2026
- **Apply:** https://www.careerbuilder.com/job-details/staff-machine-learning-engineer-ai-platform-agentic-apps-menlo-park-ca--c54c71cf-85f6-4a27-9030-a29ad84257fa

## About the Role

Machine Learning, Artificial Intelligence (AI), Finance, Problem Solving Skills, Team Building, Team Lead/Manager, Artificial Intelligence (AI), Machine Learning, Building Systems, Mentoring, * 10+ years of experience as a Machine Learning Engineer or ML-focused software engineer, with strong Python and distributed-systems fundamentals and a track record of shipping LLM-powered systems to production at scale. A Master's degree in Computer Science or a related technical field, or equivalent professional experience.

 * Hands-on experience building agentic systems end to end - tool use, orchestration, context management, multi-step planning - on top of frontier models, in production.
 * Deep expertise evaluating agents: you've built trajectory-level evals, tool-call scoring, and simulation environments, and you can articulate why final-answer accuracy is insufficient for systems that act.
 * Demonstrated expertise designing action-level guardrails - permission and tool-scoping models, approval gates, blast-radius controls, and sandboxing - for agents operating in systems where mistakes have consequences.
 * Rigor in evaluation methodology: golden datasets, rubric and LLM-as-judge grading and their failure modes, statistical significance with small N, offline-to-online metric correlation, and eval data versioning and contamination control.
 * Proven ability to build platforms, not just models: you've shipped eval, safety, or agent tooling that other engineering teams adopted, and you have the judgment to know when to build versus buy.

## Description

* Design and build the core of Robinhood's agent harness - orchestration, tool integrations, context and memory management - so one platform can safely power both high-trust internal agents and tightly scoped customer-facing ones.
 * Ship agentic applications end to end on that harness, from an ambiguous problem to a production agent that takes real action on behalf of employees or customers, and feed what you learn back into the platform.
 * Build trajectory-level evaluation systems that score how an agent got to an answer, not just the answer - tool-call correctness, planning and recovery, multi-step task completion - backed by simulation environments and synthetic task generation.
 * Architect action guardrails as platform primitives: least-privilege tool scoping, permission models, human-approval gates for high-risk or irreversible actions, step and budget limits, sandboxing, and rollback.
 * Make evals and guardrails products other teams adopt - SDKs, CI regression gates on prompt, model, and tool changes, continuous red-teaming, and production tracing that closes the loop from real traffic back into eval sets and guardrail models.
 * Set the technical bar through architecture reviews, code reviews, and mentorship, and be the person who can make - and defend with data - the "don't ship" call.