Data Analyst at UA, Ryan specializes in modeling and automation to streamline university reporting. He bridges the gap between data and business intelligence through data science and AI, leveraging ML, LLMs, and emerging GenAI to transform complex information into clear, actionable insights.
Sessions in which Ryan Ju participates
Tuesday 24 February, 2026
This talk wants to share and discuss what skills still matter in data work, the risks AI introduces newcomers, and the challenges experienced professionals face when using AI to level up their data work.
Wednesday 25 February, 2026
In this session, I’ll share how I used local large language models (Llamma, Qwen, and GPT) to automatically summarize prospect interaction records. The goal was to keep data secure while making it easier to understand and analyze. I’ll show how the models were tested and refined using both AI and human feedback, and how the final process was automated with Python and Windows Task Scheduler. The results feed directly into Power BI dashboards, turning complex text data into clear, actionable in...
Sessions in which Ryan Ju attends
Tuesday 24 February, 2026
Wednesday 25 February, 2026
This session explores how to build custom Google Gemini Gems tailored to your unique needs. You will learn how to personalize an AI assistant using persona, task, context, and output-format parameters to deliver accurate, situation-aware responses. We will also compare this recent approach with earlier technologies such as the IBM Watson Assistant used to build an Activity Recommender in Moodle LMS, to highlight improvements in flexibility and ease of integration.
I'd like to kick off a conversation on neuro-symbolic AI by sharing a personal project. In it, I build a knowledge graph using thesis metadata available as open data from the institutional repository. One of the questions it helps answer is: who has supervised theses on various topics of interest? This will be a show-and-tell session, and I hope it sparks a dialogue around symbolic AI and its potential applications at Concordia.
Learn how to create small, custom programs that solve your specific needs. No coding experience required. With AI assistance, I'll build several practical tools live, from concept to working program. Watch as we go from "I wish there was a tool for..." to a functioning program you can use immediately. You'll leave with the confidence to create your own solutions for everyday problems.
Walkthrough of Copilot, our institutional tool that ensures data security. We'll quickly explore how to access it, use it, and build with it.
If you’d rather not share your data with mysterious servers in faraway places, this session might be for you. We’ll start by clarifying what “open source” means and highlighting a few open-source LLMs. You’ll learn the key differences between locally run and cloud-based models, and why those differences matter for privacy and security. Then we’ll introduce Ollama, a simple tool that makes it easy to download small language models and run them locally. We’ll wrap up with a step-by-step demo on...
This session explores Concordia’s guidelines surrounding generative AI, including the importance of Privacy Impact Assessments (PIAs), data classification, and contractual safeguards when using platforms like Microsoft Copilot or ChatGPT. Learn practical strategies to prevent unauthorized data use, comply with Quebec’s privacy laws, and align with Concordia policies and guidelines. Gain actionable insights to balance innovation with ethical, secure practices—empowering the community to use AI...
Presentation of 3 or 4 scenarios pertaining to data privacy. Presenters will share tools and tactics to protect privacy.
