Netradyne is looking for an experienced Model Optimization Engineer to benchmark internal models and perform optimizations to achieve performance improvements, likely in throughput or memory, for their fleet safety solutions that leverage Computer Vision and Edge Computing.
Requirements
- Familiarity with the internals of various SOTA vision model architectures
- Familiarity with SOTA VLMs and an understanding of concepts involved in running inference.
- Some Familiarity with one of the Cloud Providers (AWS, Azure, etc.)
- Excellent programming skills in Python
- Experience in performance optimization of ML code (both training and inference).
- 2+ years of experience writing performant training and inference code
Responsibilities
- Work with State-of-the-Art models such as VLMs, Vision Encoders, and other model architectures to get performance improvements (most likely in the form of throughput or memory)
- Work on Edge related model optimizations considering hardware constraints
- Apply post-training model compression techniques from the latest research papers
- Perform performance optimization of ML training code to extract best performance from available hardware
Other
- Be self-driven and demonstrate the ability to derive results with limited supervision.
- Salary + eligibility for yearly bonus
- Company equity
- Company Paid Health Care, Dental, and Vision Coverage for you and most of your dependents
- Generous PTO and Sick Leave
