My name is James. My (not so anonymous) pseudonym is sanjams.
I like to consider myself an engineer; I'm interested in understanding how things work and solving problems of all shapes and sizes. Today, the form of engineering I mostly practice is software engineering. My focus here has mostly been on backend distributed systems, but I have dabbled in many other areas of the stack as well (the lower the better!). I tend to work quickly and ask questions later.
I am originally from Atlanta. I have lived in Seattle and now reside in New York City.
I studied Industrial Engineering and Computer Science at the Georgia Institute of Technology with a focus in Quality and Statistics and Networking respectively. I graduated in 2017.
I worked at AWS from 2017 to June 2025. Within AWS, I spent the majority of my time (~6 years) working on AWS SageMaker. I spent about four years on the Inference team primarily focusing on the real-time inference system which was responsible for handling millions of requests per second. Some of my projects included Model I/O Capture, Asynchronous Inference, Serverless Inference, Model Response Streaming, and performance optimizations for reducing our data-plane latency overhead to single-digit milliseconds per request. I then spent about two years on the Training team where I primarily focused on building the HyperPod compute platform. My projects included Launching HyperPod, Kubernetes support for HyperPod, and improving our fault recovery system for Nova Model Pre-Training (where we achieved 97% goodput!). I greatly enjoyed my time at AWS; I am now onto something new...
While I have built new things, I generally prefer improving things that already exist.
Hoping to eventually move these to the project section above, but feel free to take it on yourself