Hello, I’m Anmol.

Learning through research

I studied Information Technology in Jalandhar before moving to Vancouver for an MSc in Computing Science at Simon Fraser University. My earliest work was in what we would now call traditional machine learning, before deep learning became the centre of my work. During graduate school, I worked in machine learning and medical imaging, including research with the Medical Image Analysis Laboratory and Vancouver General Hospital.

It was a formative time. Medical imaging is a good place to learn that accuracy, robustness, and reproducibility are not decorative ideas. They matter because the work is connected to people and decisions that deserve care.

As deep learning changed the field, generative AI became my main graduate research focus. I worked with generative adversarial networks to synthesise missing MRI pulse sequences, alongside work in microscopy. I spent a lot of time learning the unglamorous parts of research: reading carefully, writing clearly, running the same experiment again when it matters, and understanding the limits of a result.

Taking machine learning into products

Alongside and after research, I worked on applied machine-learning problems in health, imaging, video, and language. I learned quickly that a working model is only the beginning. A useful product also needs good product judgment, reliability, a realistic view of its users, and a team that can make decisions when the path is not obvious.

At Xtract AI, then part of Patriot One, I spent an important chapter of my career in the defence and security world. We were building AI-enabled products across video processing and natural language processing, in a domain where the consequences of getting the details wrong were tangible. I grew from Senior Machine Learning Engineer to Engineering Manager and then Director of Engineering.

At BenchSci, I led work around machine learning and NLP for scientific discovery. These were the years when I began to care as much about the conditions that help people do good work as I did about the technology itself.

Working on the platform layer

Today I lead ML Workflows at Weights & Biases, now part of CoreWeave. A large part of my work has been bringing greenfield products to life, staying close to the technology while helping the company build the teams and operating shape needed to support them.

Registry was one of those stories. I helped take it from a proof of concept to general availability, and it is now used by some of our largest customers, including Pinterest, Dropbox, and Canva. I then helped build the first version of the W&B mobile app, hired the team around it, and launched it at NVIDIA GTC.

My current focus is CoreWeave Sandboxes. I am working alongside the CoreWeave platform team to take the product from private preview through general availability, while scaling the organization to support the expanding product surface area. The work spans secure execution for AI workloads, developer workflows, compute orchestration, automation, and product experience.

I still like being close to the technical details, but I am increasingly interested in the organizational work too: helping people make good decisions, turning ambiguity into direction, and building teams that can be trusted with meaningful problems. I think the next generation of AI products will be defined as much by the quality of their infrastructure and user experience as by the models behind them. That is the part of the work I am most excited about.