Founding Software Engineer

We’re looking for a senior software engineer with deep intuitions about concurrency who is comfortable writing performance-critical code across the stack, from compiled C runtimes to Python agent orchestration and TypeScript frontends. You’ll work directly with the founders on the core platform: the robot interface, the real-time world model, and the APIs that tie everything together.

The systems you build ship into customer environments. Correctness and uptime matter from day one, and you’ll own your decisions end to end.

This role is based in New York City. We work a hybrid model with a minimum of three days per week in our SoHo office.

In this role, you will

• Design and implement the distributed execution layer that coordinates robot agents, validators, and skill runners across work cells
• Own the concurrency model for the platform, from scheduling and synchronization to failure recovery
• Build the real-world representation layer that turns camera input (objects, positions, states) into a live, persistent picture the system can reason over
• Write performance-critical runtime components in C that turn design-time agent reasoning into deterministic, low-latency execution
• Develop the TypeScript interfaces and APIs that surface robot status, task state, and live feedback to operators
• Instrument the system for observability, including tracing, latency profiling, and failure analysis across a distributed fleet
• Work closely with ML and robotics engineers to keep the execution layer and the learning loop tightly coupled

• Have 5+ years of experience building distributed systems, with production exposure to concurrency bugs, partial failures, and consistency tradeoffs
• Write fluent Python and TypeScript, and are comfortable dropping into C when performance demands it
• Have strong intuitions about concurrent execution and know when to reach for locks, queues, or message passing
• Have built systems where correctness is non-negotiable, such as robotics, autonomous systems, real-time control, or similar domains
• Design for retry, rollback, and graceful degradation before the happy path
• Move fast without cutting corners on correctness, observability, or API contracts
• Are comfortable switching between a C runtime, a Python orchestrator, and a TypeScript dashboard in the same afternoon
• Have hands-on experience with large language models (LLMs), agent frameworks, or model inference, though this is a bonus rather than a requirement

Base salary: $165,000-$250,000

Mbodi builds the software layer that lets industrial robots learn new skills from instruction in minutes instead of weeks of hand-tuning. We work with Fortune 100 companies across manufacturing, logistics, and pharmaceutical labs, where we operate at the intersection of real-time robotics control, distributed agent orchestration, and compiled neural inference.

As one of our early engineers, you’ll work on the core systems that translate camera input into a software representation of the physical world, orchestrate agents that reason over it, and compile learned skills into fast, deterministic robot behavior on the fly., Mbodi makes industrial robots learn like humans. Our AI software plugs into any robot and lets anyone teach it new skills through conversation. We’re backed by leading investors. Our founders bring experience from Google and the University of Pennsylvania’s robotics lab, and we’re building a small, senior team focused on hard problems in the physical world.