Mbodi AI hires founding machine‑learning engineer as it scales embodied‑AI robotics platform

Mbodi AI – a brief overview

Mbodi AI is an embodied AI platform that enables industrial robots to acquire new skills through spoken instructions. The core idea is to let a non‑technical operator say "pick up the box and place it on the conveyor" and have the robot translate that command into a reliable motion plan within minutes. The company combines large‑scale generative models, vision‑language systems, and reinforcement‑learning pipelines to bridge the gap between foundation models and the physical world.

Founded in 2024 by Xavier (Tianhao) Chi and Sebastian Peralta, both alumni of Google and the UPenn GRASP robotics lab, Mbodi AI is part of the Y Combinator Winter 2025 batch (P25). It has already secured backing from several prominent investors (see list below) and is working with global industrial partners such as ABB and multiple Fortune‑100 manufacturers across logistics, manufacturing, and laboratory automation.

---

The problem Mbodi AI tackles

Traditional industrial automation relies on painstakingly programmed motion sequences and extensive calibration. Re‑tooling a line for a new product can take weeks, and any deviation in the environment often forces a costly shutdown. Mbodi AI’s platform aims to replace that rigidity with human‑like learning:

• Natural‑language teaching – operators can convey new tasks verbally, reducing the need for specialized programmers.
• Rapid skill generalization – the system abstracts the core intent of a command and applies it to varied objects and layouts.
• Production‑grade reliability – despite the flexibility, the platform enforces safety checks, latency budgets, and observability so that learned skills meet industrial uptime standards.

If successful, this approach could shrink the time‑to‑value for automation projects from months to days, a compelling proposition for manufacturers facing ever‑shorter product cycles.

---

Funding and market positioning

Mbodi AI disclosed a seed round of $100 K–$250 K, granting the new founding engineer a 0.5 %–2 % equity stake. The round was led by First Round Capital and included participation from Amplify Partners and several YC alumni angels. While the amount is modest, the equity range reflects the seniority of the role and the company’s early‑stage status.

The startup positions itself at the intersection of three fast‑moving markets:

1. Industrial robotics – a $70 B market projected to grow 10 % CAGR, driven by demand for flexible automation.
2. Generative AI – large language and vision models are increasingly being adapted for control tasks.
3. Enterprise AI platforms – companies are looking for turnkey solutions that hide model complexity behind simple interfaces.

By integrating these trends, Mbodi AI hopes to differentiate from pure‑hardware players (e.g., FANUC) and from software‑only orchestration layers (e.g., Robot Operating System extensions). Its early partnerships with ABB give it credibility on the hardware side, while its YC pedigree signals a strong product‑market fit engine.

---

The founding Machine Learning Engineer role

The opening is framed as a founding position, meaning the hire will join a senior team of six and help shape the core technology stack. Key responsibilities include:

• Conducting applied research that blends generative AI (transformers, diffusion models) with robot perception and planning.
• Designing algorithms for skill generalization, enabling a robot to transfer a taught behavior to new objects or environments.
• Building end‑to‑end pipelines that move models from training in the cloud to real‑time inference on edge controllers.
• Ensuring production‑grade reliability—latency under 100 ms, observability dashboards, and rigorous failure‑mode testing.
• Driving architecture decisions for the agentic AI system that coordinates perception, language understanding, and motion execution.

Technical stack: Python, PyTorch, ROS 2, and a mix of cloud services for data orchestration. Experience with vision‑language models, imitation learning, or robotic foundation models is explicitly called out.

---

Why the role matters for the broader AI‑robotics field

Embedding large generative models in physical agents is still an open research problem. Most successes to date are confined to simulation or tightly controlled lab setups. Mbodi AI’s approach—pairing a conversational interface with a production‑ready robot—offers a concrete testbed for evaluating how well foundation models handle real‑world uncertainty, safety constraints, and latency requirements.

A senior engineer who can translate cutting‑edge research into a robust, deployable system will provide valuable data points for the community. Lessons learned about model drift, sensor noise, and safety verification could inform academic work on embodied AI and guide other startups aiming to commercialize similar technology.

---

How to apply

Interested candidates should have at least three years of experience shipping ML‑driven products, be comfortable with both research and systems engineering, and be willing to work in New York (visa sponsorship is available). Applications are accepted through the company’s career page, where candidates can connect directly with the founders.

---

Mbodi AI’s progress will be a useful barometer for how quickly the industry can move from isolated AI breakthroughs to reliable, large‑scale robot deployments.