Yuval Boss

Welcome to my site! I registered this domain in 2011 and have enjoyed hosting a variety of iterations of this site over the years, starting as just a simple IRC that friends and I could use to message eachother on our school computers without the teachers knowing.

Since then building websites has become much more automated, I now host this using GitHub Pages, larger assets are hosted on s3 and served by cloudfront, and I use Jekyll for generating static web pages such as this one you are on now.

Current Work

Currently I am a software engineer at Fiddler AI where I get to work on some super interesting problems with a great team. Early on I got to build up our inference capabilities leveraging Ray, which has an incredible developer team and open source community backing it. While the inference server continues to be an ongoing focus I also have spent some time working on our core Evals offerings, on guardrails, on MCP, and on agentic evals test design. Maybe my favorite part of this work has been how new this all is, which has meant getting to work on many problems that are either unsolved (or poorly solved) with so much room for creativity and invention.

One of the most significant parts of this experience for me is that this was the job I was at when all this agentic coding stuff took off. When I started at Fiddler, I wrote 100% of my code, today I write very little code, if any. I can’t imagine being on a better team with the flexibility to embrace it all while simultaneously having to solve all the new challenges it creates. It is an unprecedented time to say the least and has been a truly wild (and for the most part enjoyable) ride.

Prior Work

Prior to Fiddler I worked at Amazon Web Services in data center supply chain automation. My team owned the tracking and availability of asset data for all AWS data center assets through their entire life. It was a wildly different experience from previous jobs, and I got to work on very interesting challenges at scales that still boggle my mind. It was a very fun experience and wild to be operating at the center of it all.

Prior to AWS I led Machine Learning development for the Polar Ecosystems Program at the Alaska Fisheries Science Center’s Marine Mammal Lab. We developed models and methods for real time detection and classification in aerial surveys, which involved a lot of R&D to develop the models and algorithms, along with MLOps and systems work to support development and deployment. Some of my favorite parts of the job were:

• Developing methods for fusing multispectral data to improve small object localization and classification.
• Semi-supervised learning approaches for large, partially labeled, and extremely sparse datasets.
• Ensemble approaches for picking up missing labels.
• Robust and reproducible data pipelines which enabled the continuous learning cycle as we trained models, collected more data through surveys, and updated models.
• Developing labeling workflows for scientific applications.
• Computer vision approaches for quantifying various image quality and other image description metrics.
• Had the honour of being invited to give a talk about this work at ICIP 2021 (slides)

Outside of work

Outside of work I enjoy the beautiful nature we have here in the PNW. In the drier months I spend most of my free time climbing at Index, Washington Pass, or on many other rocks that can be found in Washington. While in Seattle I’m often at the UW Rock, a training mecca and rendezvous for city monkeys.

Check out my trip reports to read about some of my adventures.