Kaushik Jeyaraman

I built and released Video2PdfSlides as a command-line tool that converts lecture videos into clean PDF slides using OpenCV, helping students and educators extract structured study material from long-form content. I designed and implemented the end-to-end workflow for frame extraction, slide detection, and PDF generation, then packaged it for easy real-world use. The project gained 225+ GitHub stars and has seen actual community usage, validating both its usefulness and usability.

I built a prototype transcript-aware video search system that used OCR and NLP embeddings to retrieve the most relevant moments inside long-form video content. I designed the end-to-end retrieval flow, from extracting searchable text signals to ranking timestamped results for natural-language queries. The project showcased practical applied AI for search and information access, and won 1st place in an internal hackathon.