MLTO: Machine Learning Toronto

Please note that RSVPs for this event are closed on Meetup - you must register directly at the event page at https://www.mlto.ca/scalingml to attend. Registration will close May 14th.

Note: RSVPs are closed - to attend, you must register on Lu.ma here: https://lu.ma/4856zsyt

Agenda
6:00-6:30 PM Arrivals and Networking
6:30-7:00 PM Tech Talk #1
7:00-7:30 PM Tech Talk #2
7:30-8:30 PM Networking & Wrap-up
8:30 PM Tear-down and Departures

Talk #1 - Moein Hasani, Data Scientist @ Instacart - From Matching to Generation: Rethinking Retrieval with LLMs
In this session, Moein will explore how to transform large language models into high-performance retrieval engines using cutting-edge generative techniques. Specifically, he'll focus on leveraging vLLM, a high-throughput inference engine, to efficiently retrieve articles from a library of titles in response to a user search query.
Moein will walk through the paradigm shift from traditional similarity-based information retrieval to generative information retrieval (GenIR) — where models generate the names of relevant documents directly, bypassing the need for dense vector indexes or exact keyword matches. Key topics include Generative Retrieval by Document Name, Constrained Generation for improved relevance and precision, and Scalable Inference with vLLM for fast and memory-efficient serving of generative models in retrieval settings.

Talk #2 - Brett Kennedy, Data Scientist & Author - Outlier Detection in Python
Outlier detection is a key part of data science, and frequently appears in our work, for example in tests for security, fraud, data quality, drift monitoring, forecasting, and many other areas. In this talk, Brett will introduce the ideas of outliers and outlier detection, go through the motivations for searching for anomalies in data, and will explain some of the challenges involved and solutions to these.

About the Speakers:

Moein Hasani is an AI enthusiast with years of experience as a machine learning engineer and researcher. With a Master's degree in Computer Science where his thesis focused on using large language models to predict interactions between bacteria and viruses, Moein has worked on applying AI across a variety of fields. At his previous job, he focused on solving real estate problems using machine learning — building solutions like document visual question answering, RAG-based chatbots, and product categorization for marketplace platforms. Currently at Instacart, Moein focuses on improving recommendation systems and is passionate about staying up to date with the latest in AI.

Brett Kennedy is a data scientist with over 15 years of experience in data science, and over 30 years in software development. He previously led the Research team at CaseWare International, where his team worked in several areas of machine learning, with a strong focus on outlier detection, particularly applied to identifying financial errors and fraud, as well as anomalies in text documents. Brett also worked with outlier detection extensively in social media analysis, focusing on searching for information campaigns through the detection of unusual behavior or unusual networks of users. He has continued his research in the area, contributing to several GitHub projects and authoring "Outlier Detection in Python" (Manning Publishing).

Many thanks to our supporters and sponsors!
We'd also like to acknowledge our venue partners for generously providing their amazing space for the event, and Trajectory Labs and PRAKTIKAI for making the event happen.

Speak or Support MLTO!
We are constantly looking for new speakers and sponsors to share their knowledge and help build the AI community in the GTA! If you are interested please get in touch directly or apply via one of the links below:
Speaker Application Form: mlto.ca/speak
Sponsor Inquiry Form: mlto.ca/sponsor
Volunteer Application Form: mlto.ca/volunteer