Suryansh Anand

AI engineer · RAG, agents, and the systems around them

prof_pic.jpg

I build AI systems meant to survive real use: retrieval pipelines that answer in under a second, agents that hand work to each other, and the backend, frontend and CI that keep them running.

Lately that has meant:

  • Shree Hari Vani — a scripture library and study assistant over 48 Hindu texts, on the web and Android, with a voice layer in 12 Indian languages.
  • EcoShield — an agentic RAG platform that turns 1,300+ pages of Indian environmental law into compliance steps for small factories.
  • Phoenix_37 — a court-defensible forensic pipeline for CCTV/DVR footage, built for Smart India Hackathon 2026 (NTRO).
  • Open source with GirlScript Summer of Code, and Gemini workshops and build days as a Google Student Ambassador, now in my second term.

I also spend a lot of weekends at hackathons, from ISRO’s Bharatiya Antariksh Hackathon to leading a team to 16th of 285 at Deploy or Redacted.

Outside of work I’m a Linux person. I’ve hopped from Mint to Ubuntu to Fedora to Arch, and settled on EndeavourOS for now.

research

  • Token-aware architecture — my first paper, which won a Best Paper award.
  • Energy-efficient LLM inference for Indian languages — in progress. Making models cheaper and greener to serve in low-resource languages like Odia and Bengali, without losing answer quality.

More on the projects page.

latest posts