VAM! Talks: AI for Scientific Discovery ⭐

​AI for Scientific Discovery ⭐

​This event explores how AI is being used to accelerate scientific discovery, from developing reliable ways to evaluate virtual cells to building AI-powered scientific workflows.

​The talks will explore two perspectives:

• ​🧬 Virtual Cell Benchmarking: We will explore how virtual cells are evaluated, where current metrics can fail, and how more reliable benchmarking can improve their evaluation.

• ​🔬 AI-Accelerated Science: We will explore how AI, self-improving agents, and automated research pipelines can enhance scientific workflows and accelerate discovery.

​This session is for AI engineers, researchers, scientists, founders, product builders, data scientists, students, and anyone interested in AI for science.

​🗓️ Program

​5:30 PM – 5:50 PM 👋 Arrival & networking
Doors Close at 5:45 PM
5:50 PM – 6:05 PM 🎙️ Issam Laradji | AI Agent Reliability
6:10 PM – 6:50 PM 🧬 Harshinee Sriram | SWARM-FR: Benchmarking Virtual Cell Metrics
6:50 PM – 7:00 PM ☕ Break
7:00 PM – 7:40 PM 🔬 Gaurav Sahu | AI-Accelerated Science & Building Self-Improving Agents That Work
7:40 PM – 8:00 PM 🤝 Closing & networking

​Timing is approximate and may shift slightly based on audience questions and discussion.

​ℹ️ Disclaimers

​🫧 Feel free to bring your own food and water bottle.
💧 There is a water fountain at the venue.
🚗 Paid parking is available nearby, including EasyPark.
🐾 The venue is not pet-friendly (service animals welcome).
🚲 Bicycles are not permitted inside the building.
⏰ Doors close at 5:45 PM, so please arrive before then.
🏢 Before 5:45 PM, someone in the lobby can guide you to the event space on the 14th floor.

​🎤 Speaker 1: Harshinee Sriram

​Computer Science PhD Candidate, University of British Columbia

​Harshinee Sriram is a Computer Science PhD Candidate at UBC. Her PhD research focuses on AI for Alzheimer’s disease diagnosis and drug discovery, and she has also worked on user personalization and explainable AI (XAI). Her industry research includes LLM evaluation and benchmarking, computer-use agentic workflows, and AI for autonomous knowledge discovery.

​Talk: SWARM-FR: Benchmarking Virtual Cell Metrics

​Harshinee will present SWARM-FR, a framework for benchmarking virtual cell performance metrics. The work evaluates 396 metrics for fidelity and robustness, identifies vulnerabilities across metric families, and proposes sets of complementary metrics that collectively address individual vulnerabilities for more reliable virtual cell benchmarking.

​🎤 Speaker 2: Gaurav Sahu

​CEO & Co-Founder, Unified Sciences | Postdoc, Mila – Quebec AI Institute

​Gaurav Sahu is the Co-Founder and CEO of Unified Sciences, where he is building the infrastructure for AI-accelerated science. His research spans self-improving agents, language-model reasoning, and autonomous research systems, with a focus on accelerating scientific and engineering workflows with AI. He was previously a postdoctoral researcher at Mila – Quebec AI Institute and earned his PhD in CS from the University of Waterloo. He also manages the infrastructure of leading AI conferences like ICLR and is building tools to scale peer review.

​Talk: Building Self-Improving Agents That Work

​Today’s AI systems talk really well, and are widely used in chatbots and many other applications. But can they do more than answer questions? Can they actually do research?

​In this talk, Gaurav will discuss recent work at Unified Sciences on building autonomous research systems that close this loop, and how they’ve made advances in hard problems in compute. He'll show how AI systems can accumulate knowledge through experimentation and use it to push the frontier. Finally, he’ll briefly discuss how these ideas might extend beyond software to the physical sciences and enable breakthrough discoveries.

The central message is simple: moving from AI that talks about discovery to AI that makes discoveries requires disciplined, experimentally grounded, self-improving loops.

​🎙️ Opening: Issam Laradji

​Adjunct Professor, UBC

​Opening: AI Agent Reliability

​Issam will introduce AI agent reliability, looking at how we can evaluate whether AI agents work reliably across tasks and environments, where they can fail, and what is needed to build more dependable agentic AI systems.

​🌲 About

​Vancouver AI Meetup (VAM!) is a community of researchers, engineers, founders, students, and AI practitioners who come together to learn, share ideas, and discuss the latest developments in artificial intelligence.

​Northeastern University was founded in 1898 and is a global research university and leader in experiential lifelong learning. With campuses across the U.S., U.K., and Canada, over 320,000 alumni, and more than 3,800 partners worldwide, Northeastern's global university network serves as a platform for scaling ideas, talent, and solutions.

​Located in the heart of downtown Vancouver, the Vancouver campus offers a wide range of master's-level technology programs, including pathway programs that allow students with non-STEM backgrounds to pursue graduate studies in technology.