Akshit Bhalla

Akshit builds the systems that take machine learning from research to production, with a focus on model inference, ML infrastructure, and the platforms that let teams build, evaluate, and ship generative AI applications at scale.

He designed and runs the model inference platform behind the Edit object tool in Adobe Firefly, which launched at Adobe MAX 2026. The platform gives research and product teams a single interface to more than a hundred image, video, and language models, and it now runs in production. He was also part of the team behind Project Hands On, a MAX 2026 Sneak that builds content-aware controls directly on the canvas, so people can make precise edits through direct manipulation instead of rewriting a prompt. Earlier, he contributed to Project Stardust, a generative AI-powered, object-aware image editing engine. He is a named inventor on several pending patent applications.

His research interests include ML systems and efficient inference, distributed training and inference, LLM agents and tool use, evaluation and observability of generative models, and multimodal models for images and video. He collaborates with university researchers, including Prof. Heng Ji’s group at UIUC, and mentors Adobe Research interns. He earned a Master’s degree in computer science from Carnegie Mellon University. More details can be found on his LinkedIn.