Senior Software Engineer, AI Platform

About the role

Dexmate is building the foundational layer for physical AI, a unified platform that tightly integrates high-quality robotic hardware with a universal Physical AI OS so that robots can be built and deployed as easily as software. In this role, you will own the design, implementation, and end-to-end delivery of AI-powered features that span agentic infrastructure, backend services, and user interfaces. You will be responsible for making agents reliable in production by deeply understanding and mitigating failure modes such as non-determinism, prompt injection, runaway tool-calling, and token cost spirals. You will architect and build the harness that turns the model into a component of a larger, production-grade system rather than treating the model as the entire product. This position is a core part of our mission to democratize robotics by lowering the barrier to entry for developers, researchers, and enterprises. If you want to help shape the next layer of human capability and believe the future of robotics should be built collaboratively, we want to build it with you.

Key facts

What you'll do

You will design, implement, and deploy production-grade AI agents that handle multi-step reasoning, tool-calling workflows, multi-agent coordination, and human-in-the-loop handoffs. You will create and maintain the agent harness, which is the runtime infrastructure responsible for context management, tool definitions, memory, feedback loops, observability, and lifecycle control to ensure agents are reliable in production. You will engineer context pipelines that perform dynamic retrieval, re-ranking, semantic search, and GraphRAG as active tools within an agentic reasoning loop, understanding when to use long context, agent memory, or retrieval. You will implement robust production-grade reliability patterns such as retry logic with backoff, cost controls, structured output validation, sandboxed tool execution, and checkpoint-resume for long-running agent workflows. You will develop systematic evaluation frameworks including evals, golden datasets, regression suites, and observability traces that measure agent quality and catch regressions before they reach production. You will architect and implement scalable backend services and APIs using Go, Rust, or TypeScript/Node.js, focusing on clean interface design, performance, and maintainability. You will build and maintain integrations with external systems such as databases, internal APIs, and robot data streams to enable agents to take real actions with appropriate access controls. You will own deployment, monitoring, and observability for these services using Docker, Kubernetes, CI/CD pipelines, and LLM-specific tracing and cost tracking. You will build clean, functional web interfaces in React and Next.js, including operator dashboards for robot fleet management, engineering tooling for the AI team, and customer-facing applications. You will own features end-to-end, from gathering product requirements through implementation, testing, rollout, and ongoing maintenance and iteration. You will treat prompt engineering as a first-class engineering discipline, writing, testing, and versioning prompts with the same rigor and discipline as application code.

Requirements

You must have 5 or more years of professional software engineering experience with a full-stack production track record, demonstrating strong fundamentals in system design, data structures, algorithms, and code quality. You must possess a strong command of Python and/or TypeScript at a production level, writing clean abstractions, testable code, and performance-aware, maintainable systems. You must have backend engineering depth in Go, Rust, or TypeScript/Node.js for production services, including RESTful and GraphQL API design, relational database modeling with PostgreSQL, async programming, caching, and system integration via APIs and webhooks. You must have frontend engineering proficiency in React and Next.js with TypeScript, capable of architecting and shipping functional, production-grade UIs rather than merely wiring component libraries. You must follow software delivery practices such as automated testing across unit, integration, and end-to-end tests, CI/CD pipelines, structured code review, and robust observability including logging, metrics, and alerting. You must have hands-on experience with containerization and deployment using Docker and Kubernetes, able to own a service from code through production without requiring a DevOps handoff. You must have proven, hands-on experience building and deploying LLM-powered systems or AI agents in production environments.

Nice to have

Preferred items from SOURCE are not specified in this listing.

Practical notes

This role is full-time and based in the Fremont Office. No specific hours, travel requirements, visa details, or application deadlines are stated in the provided SOURCE text.