​Curious about taking your LLM applications beyond basic prompts? Join us for an exploration into the world of structured outputs and open source development. We'll dive into how simple yet powerful techniques like function calling can transform the way we build with LLMs, making them more reliable and easier to integrate into existing systems.

​More about the Sharings

• ​Ivan Leo (Research Engineer, 567 Labs) will share insights from his journey in open source development and how simple function calling can unlock unexpected possibilities in LLM applications. The talk will showcase practical applications through his work on popular open source libraries including Instructor (Python object validation for LLMs) and Kura (user data pattern visualization). (Technical Difficulty: 200)

• ​Yichao Jin (Data Scientist, GovTech) from GovTech Digital Economy Product division will share their technical journey on building an on-prem hostable In-house Government Document Understanding Service that can process sensitive-high scanned/photo-taken documents. The sharing will cover the end-to-end pipeline implementation details on image preprocessing, document classification models, local OCR engine, key information extraction models, and post-processings. The services had been integrated into MOH's foreign nurse registration system, effectively simplifying the document verification process and shortening the approval period. The team is in the midst of expanding the same service to other agencies. (Technical Difficulty: 200)

​More about the Speakers

• ​Ivan Leo is a Research Engineer at 567 Labs, where he focuses on synthetic data generation and evaluations for LLMs. He is an active open source contributor and maintains several widely-used libraries including Instructor, Kura, and Indomee. His work centers on making LLM implementations more robust and accessible to developers.

• ​Yichao Jin is a lead data scientist at GovTech, with a PhD degree in Machine Learning and AI from NTU obtained in 2015. He authored over 30 journal and conference publications in this research area, and his decade of industry experience served as a trusted guide for applying AI to a variety of real-world products.

​Do note that the official sharing begins at 3pm. Registration and Networking begins at 2.30pm.

​NOTE: Attendance is prioritized for Lorong AI members.

​About Lorong AI

​Lorong AI is a co-working hub where AI practitioners connect, share knowledge, and grow through curated programming and a collaborative environment. Home to programmes like AI Wednesdays and AI ToolsDays, Lorong AI offers hands-on workshops, technical deep dives, and opportunities to solve real-world challenges with AI.