Sankalp Tewari
Based in IndiaEngineer of AI products that ship.
I build AI systems for production—quiet, confident, with visible machinery. Every claim here is backed by an artifact: a working demo, a strict latency figure, or a rigorous eval score.
Core Technologies
Frontend
- React
- Streamlit
- TypeScript
- Next.js
Backend
- FastAPI
- Node.js
- PostgreSQL
- Redis
AI / ML
- PyTorch
- Hugging Face
- FAISS
- LangChain
DevOps
- Docker
- Vercel
- CI / CD
- Linux
Selected Works
Zero-cost, real-time deepfake audio detection intercepting live WebRTC/Twilio streams via a decoupled multi-model CPU pipeline.
A secure, natural-language semantic search engine built on an untrusted LLM extractor, relying on a strict 5-stage validation pipeline.
Awards & Recognition
Winner — SIH 2026 Internal
Won the internal SIH 2026 selection hackathon with VoiceTrace, an AI-powered voice-cloning impersonation detection system.
1st Runner-Up — Hack-e-Awadh
Secured 2nd place out of numerous teams with Commudle Sense, a semantic search system for natural-language discovery across developer communities.
About Me
Architecting intelligence for the real world.
I'm Sankalp Tewari — a Computer Science undergraduate and AI Engineer with hands-on experience across applied machine learning, backend architectures, and scalable web development. I build production-ready AI/ML solutions using PyTorch, LangChain, and modern web frameworks, always prioritizing performance, strict latency constraints, and a rigorous, detail-oriented mindset.
