New York AI, ML, and Computer Vision Meetup

With the growing ubiquity of GenAI use-cases and deployment across industries, new vulnerabilities and risks are emerging that are not addressed by traditional cyber security or software development guardrails. This session will review examples of new Large-Language-Model (LLM) and Computer Vision use cases in a number of industries (financial services, telco, automotive, and more), the new vulnerabilities they create, and ways to mitigate such new threats and risks.

Financial markets move at the speed of information, and traders are increasingly looking beyond traditional text-based data for an edge. Vision-Language Models (VLMs) offer a revolutionary approach by analyzing both visual and textual financial data – from satellite imagery and corporate filings to stock charts and social media trends. This talk explores how VLMs can extract alternative trading signals, interpret financial reports beyond text, and analyze sentiment from news images, all in real time. We’ll also discuss the challenges of accuracy, bias, and regulation when applying VLMs to financial decision-making. By integrating multimodal AI into trading strategies, investors can see the markets like never before.

As data-centric AI continues to grow, more time is spent curating data for better training outcomes. Over the last few years, the script has been flipped. Nowadays, models are used to inform data curation, creating a feedback loop of incremental improvements to data and models side-by-side. We’ll review some literature from the last few years, and see how we can use FiftyOne to quickly start model training on any FiftyOne dataset.