Abstract
Coding models are helping software developers move even faster than ever before, but weirdly, they’re not keeping up with our fast progress. The models that power code generation are often based on months to years old snapshots of open source code. And they’ve never seen your in-house codebase, let alone the code you wrote yesterday. The result is a world where you, the software developer, are left with a constant need to bring your coding agents up to speed. And the common tools and workflows teams use to bring these models up to speed are all pretty imperfect, leaving developers constantly playing catch up, when they should be focusing on the new features to be developed. This talk will introduce you to the emerging workflows to use open source tools to bring your coding agents up to speed, using efficient, frequent fine-tuning of coding models. We walk through how you can extend from the generalist coding intelligence capabilities and build internal tools that are regularly being proven to solve problems for your team.
Speaker
Jeff Smith
CEO & Co-Founder @ 2nd Set AI, AI Engineer, Researcher, Author, Ex-Meta/FAIR
Jeff Smith is an AI engineer, researcher, and entrepreneur. As the Co-Founder and CEO of 2nd Set AI, he’s working on the future of reasoning in generative AI. Previously, he was at Facebook/Meta, leading some of their most important AI initiatives including PyTorch, various fundamental methods breakthroughs in FAIR, and the productization of EMG-based neural wristbands. Prior to that, he had a long career across a range of AI and biotech startups, across the US, Asia, and Europe. He’s the author of technical books such as Machine Learning Systems from Manning, and he regularly talks about his work at conferences. His research focuses on efficient model learning methods and has resulted in the development of methods such as SHARe-KANs.
