Computer Science undergrad who ships full-stack, AI-integrated products — not tutorial clones.
Two deployed, end-to-end AI systems below — real users, real deployments, no hallucinated citations.
I'm a CS student at REVA University (Semester IV), and my repos are mostly built to solve a real, specific problem end-to-end rather than to check a resume box — auth, database, error handling, deployment, and all. Two of the projects below are live and deployed, not just localhost demos.
I care about grounded AI — systems that cite real data instead of hallucinating, and that fail gracefully when a model provider goes down. That shows up in both flagship projects below.
|
🏛️ LegalMinds AI legal analysis for Indian law that doesn't hallucinate statutes. Turns a plain-English description of a legal problem into a structured analysis — applicable laws, rights, penalties, timelines, costs, filing portals — with every citation matched against a curated database of 60+ real Indian statutes (IPC, CrPC, Consumer Protection Act, IT Act, and more). The model is constrained to that list, so it can't invent section numbers. Also generates court-submittable legal notices and exportable PDF reports, bilingual (English/Hindi).
|
🛰️ InfraWatch Satellite-based illegal-construction detection for Indian municipal corporations. Built for Google Solution Challenge 2026. Feeds satellite tiles through Gemini 2.5 Flash to flag unauthorized construction, auto-assigns an SLA tier (4–24 hrs based on confidence), routes the case to a field officer, and drafts a legal notice citing real municipal statutes — falling back to a generic reference rather than inventing a section if it can't confidently match one. Includes automatic failover to a second AI provider if Gemini rate-limits, plus a public citizen-reporting portal.
|
Languages
Frontend
Backend & Data
AI
Tools
📫 Reach me at [your-email@example.com] · Open to internship and placement opportunities in full-stack and AI/ML engineering.