Machine Learning Engineer - Neural Network Optimization

Remote /
Research /
Full-Time
Fyusion is a leading machine learning and computer vision company focused on automotive inspections and related applications. Our patented 3D format enables anyone to capture and display interactive 3D images using their smartphone, and enables significant added functionality with deep visual understanding and machine learning-driven analysis.

Founded in 2014, Fyusion is now part of the Cox Automotive family. Our team includes some of the world's top researchers and developers in light field imaging and AI, continuing to push boundaries and innovate at the highest level from our San Francisco research center.

The Machine Learning Team at Fyusion designs, creates and maintains the neural networks that power our intelligent data pipelines. This team is integral to delivering the intelligent imaging products beloved by our customers.

This role will work with our best-in-class research and product teams, taking an active part in improving the performance of vehicle smart imaging and damage detection pipelines. As part of this effort, you will prototype, iterate on, deploy, track, optimize and scale our neural networks, handling thousands of images every minute. This is a high-impact role that provides many opportunities for learning and professional development for highly motivated self starters.

Here's the day-to-day:

    • Optimize current and develop new efficient inference algorithms.
    • Distill/Quantize/Prune current architectures and/or propose new more efficient architectures. 
    • Demonstrate robust exploratory/experimental skills. 
    • Propose and test hypotheses for each new experiment, document the findings.
    • Maintain insight into trends and advances in machine learning, particularly in fields of Distillation, Quantization and Pruning.
    • Implement and advance models from the research community, including literature surveys and implementing novel techniques from research papers.
    • Work with different teams to combine information from several sources and develop an experience that is better across the board.
    • Work as part of a team.
    • Participating in the engineering life-cycle at Fyusion, including designing systems, writing production code, conducting code reviews and working alongside our design, engineering, and QA teams.
    • Publish and present, formally and informally, on behalf of Fyusion.

Here's what we are looking for:

    • 2+ years of Python expertise (PyTorch, OpenCV)
    • Rapid prototyping experience with Jupyter Lab/Notebooks.
    • Proven code profiling and optimization skills.
    • Experience with Distillation, Quantization and/or pruning.
    • Experience with preparing and running neural networks on mobile devices.
    • Strong critical thinking and hypothesis testing abilities.
    • This position is suited for a wide range of experience levels. If this peaks your interest and you see yourself described here, please do not hesitate to apply!

Bonus points for:

    • Experience with C++.
    • Publications in the areas of Distillation, Quantization, Pruning or similar.
    • Experience with NVIDIA’s TensorRT, DALI, NSight tools.
    • Experience with hyperparameter sweeps and understanding of the Pareto front.
Here's what we can offer you:

A competitive compensation, health, vision and dental benefits with premiums paid by Fyusion, generous PTO plan, company holidays (including your birthday), and the chance to be part of a pioneering technology team!

We offer some amazing perks for those working from our SF HQ: commuter benefits, company catered lunches, a fully stocked snack pantry, tons of company off-sites, and a pup friendly workplace.

If you read this job description and saw your name all over this, apply! If you read this, and think that you might need some help hitting all of the points, please apply! We have an entire team who is happy to help and share our knowledge with you.

The benefits do not apply to contract or internship positions.