Founder at Aravi Tech Labs. I build AI tools that make images and video.
IIT Delhi '24, based in Gurgaon. I design the product and write the code, so most things here started as something that annoyed me and ended up as either a company or a repo.
Currently: JewelAI, AI product photography for jewellers — zero to 2,000+ signups and 50+ paying customers in 45 days, organic, no ad spend. I also make AI launch videos for startups, including Lanesurf (YC S25).
Founder Book — Ask 1,500+ Y Combinator talks and Paul Graham essays a question in plain English, get a cited answer. A self-hostable LLM wiki that keeps itself up to date. Python · Gemini · RAG
Thumprint — Invisible watermarking for AI images and video. 82 ms embed, imperceptible at 46–50 dB PSNR, survives re-encode and crop. Built for EU AI Act Article 50 provenance. Python · computer vision
PosterBoy — Seven agents plan, compose and critique a poster until it's production-ready. TypeScript · Gemini · multi-agent
NodeWow — Decompose a multi-scene prompt on a node graph, then generate every frame in parallel. TypeScript · React Flow · BullMQ
Veo Prompt ML — Fine-tuned Flan-T5, FAISS semantic search and a reward model for ranking video prompts. PyTorch · PEFT · FastAPI
Python TypeScript Swift · Gemini Imagen Veo Claude · FastAPI Next.js SwiftUI · Figma


