Events
Machine Learning Conclave
Our annual flagship gathering — talks, panels, and hands-on sessions bringing together researchers, practitioners, and students to explore the frontier of machine learning.
Learn MoreOther Events
Variational Autoencoders
A whiteboard discussion on Variational Autoencoders, covering the math behind auto-encoding variational Bayes, the reparameterization trick, and how VAEs learn a smooth latent space for generative modeling.
Tunequest
A finetuning hackathon focused on Small Language Models (SLMs), where teams finetune and evaluate models on code snippets across languages like Java, C++, and more. Participants work with real codebases, iterate on training runs, and are judged on both benchmark performance and the practical usefulness of their finetuned models.
LoRA and QLoRA
A deep-dive discussion on Low-Rank Adaptation (LoRA) and its quantized successor QLoRA. We walk through the core ideas behind both papers, why low-rank updates work for efficient finetuning, how quantization pushes memory savings even further, and where each approach breaks down in practice.
Techevince
Our annual project showcase, where teams demo the work they've built over the year to the wider college community. It's a chance to see ideas go from concept to working systems, with live demos, project walkthroughs, and feedback from faculty, alumni, and peers across departments.
Agents 101
An introduction to building AI agents, covering LangGraph, LangChain, and Retrieval-Augmented Generation (RAG). Participants learn how to chain tools, manage agent state, and ground responses in external knowledge to build practical, working agentic systems.
PyTorch Workshop
A hands-on introduction to PyTorch, covering tensors, autograd, and building and training neural networks from scratch. Participants write and run real training loops, learning the practical foundations needed for any deep learning project.
CLIP
A discussion on CLIP and the vision backbones behind it, including ResNets. We walk through how CLIP learns joint image-text representations through contrastive pretraining, and why that approach transfers so well to a wide range of downstream vision tasks.
Self-Adaptive Language Models
A discussion on self-adaptive language models and how reinforcement learning can be used to let models adjust their own weights or behavior at inference time. We cover the core mechanisms behind self-adaptation and what it means for the future of continually learning systems.