Back to jobs

Software Engineer – ML Robotics Systems

Acceler8 Talent
San Francisco, CA
Full-time
20,000,000 – 25,500,000 / year
AI tools:
PyTorch
Applications go directly to the hiring team

Full Description

Software Engineer – ML Robotics Systems

We’re partnered with a well-funded, deeply technical team building next-gen AI systems that operate in the physical world.

They’re hiring engineers to work across the full stack of ML-powered robotics — from large-scale data pipelines to deploying models on real robots.

👉 This isn’t siloed.

You’ll move between infra, ML, and real-world testing — and see your work come to life outside simulation.

💡 What Makes This Role Different

On any given week, you might:

* Extend a multimodal (video-heavy) data pipeline

* Improve ML training or data infrastructure

* Run experiments to validate system changes

* Deploy and test directly on physical robots

If you want to build systems that interact with the real world, this is that role.

⚙️ What You’ll Work On

* Build and scale distributed Python systems for ML workflows

* Design pipelines for massive multimodal datasets

* Improve infrastructure for large-scale model training

* Deploy and validate systems on real robotic platforms

* Drive performance and scalability as systems grow 10×+

🧠 What You’ll Own

* End-to-end systems impacting robot behaviour and capability

* Core infrastructure for data and model pipelines

* Logic that determines how robots learn from data

* Evolution of the platform as it scales toward more general AI

🔍 What They’re Looking For

* Experience building distributed systems or data pipelines at scale

* Strong Python + backend/infrastructure fundamentals

* Experience working with large datasets (ideally video/multimodal)

* Understanding of modern ML training workflows

* Cloud experience (e.g. Kubernetes, networking, permissions)

➕ Nice to Have

* Experience at very large scale (e.g. petabyte-level data)

* Exposure to robotics, perception, or embodied AI

Applications go to the hiring team directly