Role Overview
We’re looking for an AI Engineer who wants to build deep learning systems that work in the real world - not in a notebook or benchmark, but on actual hardware making thousands of decisions a minute on the messiest data you’ve ever seen. Think: a conveyor belt of crushed cans, greasy pizza boxes, shredded paper, and unidentifiable plastic, moving at full speed under variable lighting - plus a whole plant of equipment we can also see, understand, and optimize.
You’ll join a small, senior team of AI, robotics, and software engineers - close enough that you’ll learn fast and have real ownership, structured enough that you won’t be on an island. Your first few months will focus on improving and extending our object detection and classification models in production. Over the following year, you’ll touch most of the ML stack: production deployment on edge hardware, drift monitoring and retraining loops, and our growing analytics layer for throughput, belt utilization, and equipment health. Few roles give you that breadth this fast.
If you’ve ever wanted your models to do something tangible in the physical world - and to help build a platform that can flex across industries - this is the role.
What You'll Do
- Design and optimize deep learning models for real-time object detection and classification on streaming visual data.
- Build perception models that go beyond sorting - extracting operational signal from video streams to power throughput analytics, equipment monitoring, and predictive maintenance.
- Work across a wide problem surface: object detection, classification, anomaly detection, and time-series analysis on visual data. Once we’re in a facility, there’s a lot to learn from what we see.
- Work closely with our robotics and software teams to integrate models into RecycleOS and the robots that depend on them.
- Make perception systems robust to the chaotic reality of industrial environments - and build them to generalize as we expand into new materials and verticals.
- Monitor models in production, catch drift early, and design smart retraining loops to keep performance sharp.
- Squeeze every millisecond out of inference by optimizing for edge hardware running on-site.
What We’re Looking For
- Bachelor’s or Master’s in Computer Science, Data Science, Electrical Engineering, or something related.
- 2–5 years building or shipping ML models (professional experience or solid project work). We care more about how you think than how long you’ve been doing it.
- Strong Python skills and hands-on experience with PyTorch or TensorFlow.
- Solid computer vision foundations and comfort with OpenCV.
Bonus Points For
- Hands-on experience with NVIDIA Jetson or TensorRT optimization.
- Familiarity with MLOps tooling like Weights & Biases for experiment tracking and model versioning.
- Exposure to anomaly detection, time-series modeling, or operational analytics on streaming data.
- Background in ROS or integrating AI into industrial automation.
- A self-starter mindset - you like ambiguity, ownership, and moving fast.
