June 26 - Boston AI, ML, and Computer Vision Meetup

Pre-registration is mandatory to clear building security

When and Where

June 26, 2025 | 5:00 – 8:00 PM
Microsoft Research Lab – New England (NERD) at MIT
Deborah Sampson Conference Room
One Memorial Drive, Cambridge, MA, 02142

Resilient Object Perception for Robotics

A broad array of applications, ranging from search and rescue to self-driving vehicles, require robots to perceive and understand the geometry of objects in the environment. Object perception needs to reliably work in a variety of scenarios and preserve a desired level of performance in the face of outliers and shifts from the training domain. Obtaining such a level of performance requires robust estimation algorithms that are able to identify and reject outliers, as well as techniques to continually improve performance of learning-based perception modules during test-time. In this talk, I discuss my three projects on this topic: (1) solvers and a graph-theoretic framework that together help achieve state-of-the-art pose estimation performance even under high outlier rates, (2) self-supervised object pose estimators that can improve performance during test-time with accuracy comparable to state-of-the-art supervised methods and (3) a test-time adaptation method for both object shape reconstruction and pose estimation without the need for CAD models.

About the Speaker

Jingnan‘s research focuses on unsupervised learning, robust estimation and robotic perception systems. He has published papers in TRO, ICRA, IROS, RSS and CVPR. He is an RSS best paper award finalist. His open source repos have been used in both industry and academia, including one of the fastest open-source libraries for point cloud registration. His blog posts have reached the Hacker News front page multiple times. He is a co-founder of a robotics startup currently in stealth mode.

Pixie: Building a Local ChatGPT Alternative using Ollama

I built Pixie out of a desire to replace my ChatGPT workflows with a local alternative. This project runs parallel to projects like LLMStudio, caters more towards Ollama models, and thus allows us to optimize for the user experience for Ollama workflows. I will go through the different philosophies at play here, design choices and how to create a system that can substitute for the “ChatGPT experience” while remaining local and open source.

About the Speaker

Suprateem Banerjee is AI Engineer trying to solve the world. I like working in fast-paced teams working on crafting novel solutions to interesting problems surrounding unstructured sensory data. I train cutting-edge predictive and generative models surrounding Computer Vision, Language, and Audio (Digital Signals)

You Can’t Do AI Without Quality APIs

The Agentic Era is here — and it runs on APIs. In today’s AI revolution, success isn’t about who has the biggest model, but who builds the highest-quality, AI-ready APIs.
From powering intelligent agents to enabling seamless orchestration, APIs are the backbone of modern AI systems. At Postman, we see how collaboration, testing, and documentation are essential to delivering APIs that truly support AI innovation. This talk explores why robust APIs are the foundation of the AI future—because in this new era, you simply can’t do AI without APIs.

About the Speaker

Pooja Mistry (@poojamakes) is a Developer Advocate at Postman, an API platform empowering developers to build, test, and collaborate on APIs. As an AI and API content creator, storyteller, and keynote speaker, Pooja brings technical concepts to life — making them accessible, practical, and engaging for developers of all levels.

She’s passionate about helping new technologists build confidence, embrace emerging technologies, and find their place in the ever-evolving tech landscape. When she’s not sharing knowledge on stage or online, you’ll find her exploring the depths as an avid scuba diver.