OCBC x Data Science SG

For the coming May DataScienceSG Meetup, we're going inside the enterprise to see how production LLM systems actually get built. Join us as OCBC's Group Data Office, AI Lab shares two sessions—on evaluating LLM outputs at scale, and on multi-agent pipelines automating financial research on the Global Markets desk.

Community partnership with Lorong AI, and Raffles Venture Partners!

​Lorong AI brings together government, industry, and research to build Singapore’s AI ecosystem. A collaborative hub for AI practitioners connect, innovate, and shape the future of AI in Singapore throughout curated programs and a supportive community.

Agenda

6:30 PM – 7:00 PM Registration & Networking 7:00 PM – 7:30 PM LLM-evaluator7:30 PM – 8:00 PM GM Capture: An AI-Powered Multi-Agent Pipeline for Automated Financial Research8:00 PM – 8:30 PM Q&A and Closing

Synopsis

Session 1: LLM-evaluator Speaker: Rionaldi Chandraseta

Rionaldi Chandraseta will be sharing how the team built "LLM-evaluator", an evaluation harness for scoring LLM outputs. The session will cover the motivation behind the tool, how it works across multiple evaluation methods, how it can be extended with custom evaluators for different domains, and how to integrate it into real pipelines.

Session 2: GM Capture: An AI-Powered Multi-Agent Pipeline for Automated Financial Research Speakers: Praveen Prashant & Akshay Sachdeva

Praveen and Akshay will be sharing how the team built GM Capture, an AI-powered multi-agent pipeline for automating financial research including credit analysis and equity evaluation. The session will cover the motivation behind the system, how it decomposes the research process into three functionally distinct layers, how the multi-agent framework is evaluated, and how the deployment is in a production environment at OCBC's Global Markets desk.

Speaker Bios

Rionaldi Chandraseta is a Data Scientist in OCBC Group Data Office, AI Lab. With several years of experience building production-grade AI systems ranging from traditional ML models to Agentic AI, he currently specializes in working with LLM models and creating Agentic AI solutions for enterprise clients.Praveen Prashant is a Data Scientist at OCBC's Group Data Office, AI Lab, with nearly a decade of experience in the machine learning industry. He specialises in translating ambitious business requirements into tangible, scalable ML products that generate measurable revenue impact. In his current role at OCBC, he owns AI/ML products across the Global Markets desk, focusing on architecting and deploying large language model (LLM) based systems, retrieval-augmented generation (RAG) pipelines, and time series models that help business teams automate research and drive revenue. Prior to OCBC, he worked on causal inference at Gojek and co-founded a Y Combinator-backed technology startup. He holds a Bachelor's degree in Engineering Physics from IIT Madras.Akshay Sachdeva is a Data Scientist in OCBC's Group Data Office, AI Lab, where he specializes in building production-grade LLM and multi-agent systems for enterprise applications. His current work focuses on designing and deploying multi-agent AI architectures for automation, reasoning, and decision-support use cases, alongside developing stock marker prediction systems using time-series models and scalable ML pipelines. Akshay bridges research and production by translating advanced AI capabilities into measurable business outcomes. He holds a Master of Engineering from the National University of Singapore.