A benchmark and dashboard that checks whether AI brand recommendations are correct, not just present, using Amazon's ESCI relevance judgements as ground truth.

Open-source tool that measures how brands show up across ChatGPT, Claude, Perplexity, and Gemini, reporting share of voice, sentiment, and cited sources.

A marketing agent that runs a launch end to end, drafting, publishing, monitoring, replying, and reporting, and tags every post with schema.org data so AI answer engines can cite it.

Human-in-the-loop, multi-agent system for NYU SPS admissions that automates routine follow-up and reminders while keeping staff in control of judgment calls.

Automates daily competitor research across Reddit, Hacker News, and the web into a Claude-written email digest. Runs on a schedule and scales to zero on AWS.

AI IT self-service portal on AWS and Claude that auto-approves or escalates software-access requests by policy, with a natural-language chat interface.

Compares an LLM against a traditional ML pipeline on customer-review sentiment, benchmarking accuracy, auditability, and handling of sarcasm.

Statistical analysis of U.S. honey production, exploring volume, pricing, and inventory trends to surface insights for commodity forecasting.