Abu Dhabi Machine Learning Season 6 Episode 1
25.09.2025 - Abu Dhabi Machine Learning - ~3 Minutes
When?
- Thursday, September 25, 2025 from 6:00 PM to 9:00 PM (Abu Dhabi Time)
Where?
- Colnago caffe Colnago Hudayriyat Island Abu Dhabi United Arab Emirates
The Meetup page of the event:
Abu Dhabi Machine Learning Meetup Season 6 Episode 1
Thank you ADIA Lab for sponsoring the event.
Programme:
Talk 1: 2nd Place Solution for Kaggle Competition: Jane Street Real-Time Market Data Forecasting
Abstract: This talk presents the end-to-end approach behind my 2nd place solution in the Kaggle competition Jane Street Real-Time Market Data Forecasting. The dataset and setting closely reflected the realities of building models for modern financial markets. I will share the full pipeline and highlight interesting findings from developing models under these conditions.
Speaker: Patrick Yam is a Quantitative Researcher and a Kaggle Competitions Grandmaster. He holds master’s degrees in Data Science and Analytics (Cardiff University, UK) and Transportation Engineering (University of Hong Kong). Patrick specializes in machine learning for financial forecasting, with expertise in time-series modeling, deep learning, and real-time prediction systems. He has earned multiple gold and silver medals in Kaggle competitions.
Talk 2: From Lab to Road: Challenges in Autonomous Driving Perception
Abstract: Machine learning has driven significant progress in autonomous driving perception, from object detection and tracking to sensor fusion and prediction. Yet, models that excel on benchmarks often fall short in real-world conditions. This talk explores the research-to-product gap in autonomous driving perception. We’ll cover the perception pipeline (sensors, detection, tracking, fusion) and highlight challenges including domain shift, calibration, simulation-to-reality gaps, and the trade-offs between accuracy and real-time performance. The session is designed to give attendees both a high-level understanding of autonomous perception and practical insights into why the “last 10%” of the problem is often the hardest.
Speaker: Murad Smreteab is a Research Engineer at Khalifa University’s Autonomous Vehicles Lab (AVLab). He holds a B.Sc. in Computer Engineering (2021) and an M.Sc. in Computer Science (2023), both from Khalifa University. His research focuses on perception and prediction for autonomous driving, with hands-on experience in LiDAR–camera fusion, 3D detection, multi-object tracking, and trajectory prediction.
Talk 3: FinChain and Beyond: Towards Transparent Financial Reasoning in AI
Abstract: How can we build AI systems that reason through financial problems in a transparent and auditable way? In this talk, I will introduce FinChain, a new symbolic benchmark that evaluates step-by-step reasoning capabilities of large language models in financial scenarios. Covering 54 diverse topics — from compound interest tax calculations to cash flow statement analysis — FinChain provides a comprehensive testbed for symbolic reasoning in high-stakes financial contexts. By decomposing complex tasks into structured reasoning chains, it sheds light on how models “think” — and where they fail. I will also outline future directions for developing reasoning-centered financial models that align better with human expectations and real-world decision-making needs.
Speaker: Zhuohan Xie is a Postdoctoral Researcher at MBZUAI, working with Prof. Preslav Nakov. His research focuses on reasoning in large language models, with applications in financial AI and fact-checking. His work has appeared at top NLP venues including ACL, EMNLP, and NAACL, and he is currently leading new initiatives on financial reasoning and multilingual AI. Zhuohan received his PhD in NLP from the University of Melbourne in December 2024, where his dissertation focused on automatic story generation and evaluation with LLMs.