Tanish Anand

Founder · Developer · Researcher

Building at the intersection of AI, hardware, and the web.

About Tanish Anand

I'm Tanish Anand, a builder, developer, and researcher currently at Delhi Public School (DPS) Azaad Nagar. I'm the co-founder and CTO of Vivacity, a Research Fellow at IIT Kanpur under Prof. Adithya Vadapalli, and the developer of Project Rudra.

Vivacity is near-real-time video infrastructure for LLMs: an API-first pipeline that turns prompts, documents, and AI answers into mathematically exact, narrated explainer videos.

Key Achievements

#1 across 3,500+: India's Top Young Founder of the Year 2026 at Uniform2Unicorn (won ₹1,00,000 + ₹10,00,000 in credits)
Research Fellow at IIT Kanpur under Prof. Adithya Vadapalli (CSE Department)
Selected for Y Combinator Startup School
Ranked in top 20 builders across India out of 20,000+ at VIBECON
Pitched real-time computer vision drowsiness detection at HACKSHODH 2026
Independently disclosed critical API security vulnerabilities with bounties awarded
Won 15+ hackathons across North India
Built decentralized local LLM assistants using quantized Phi-3 Mini models

Projects by Tanish Anand

Vivacity (2026)

Near-real-time video infrastructure for LLMs, turning prompts, documents, and AI answers into mathematically exact, narrated explainer videos.

AEGIS (2026)

A terminal user interface for Palantir's AIP control plane, built as a focused systems project.

Project Rudra (2026)

GPU-accelerated global OSINT command grid. Real-time flight tracking, marine channels, CCTV, seismic activity in 60fps WebGL. Live at osirisai.live

Text Diffusion vs. Auto-Regressive Models (2026)

Research benchmarking discrete and continuous text diffusion architectures (SEDD, LLaDA) against auto-regressive LLMs on Hindi NLP pipelines.

EduCore (2025)

Solar-powered offline adaptive AI learning device for rural Indian students. Runs quantized Phi-3 Mini on Raspberry Pi.

PhysicsGPT (2025)

Offline AI tutor fine-tuned on CBSE Physics curriculum, deployable on edge hardware.

ComplianceGuard (2025)

Dual-LLM autonomous compliance agent. Grok handles fast scanning, Claude performs deep audit reasoning for SOC 2, HIPAA, GDPR, ISO 27001.

Anti-Sleep Pilot (2026)

Real-time drowsiness detection using computer vision. Built and pitched at HACKSHODH 2026.

Technical Skills

AI & ML: Discrete Text Diffusion (SEDD, LLaDA), Model Quantization (llama.cpp), PyTorch, LLM Multi-Agent Systems, Computer Vision, OpenCV, Manim, Celery/Redis

Web & Systems: Next.js, React 19, Tailwind CSS v4, WebGL, MapLibre, FastAPI, Render, Vercel

Hardware: STM32, Raspberry Pi Clusters, Solar Edge Computing, Quantized Local Inference

Security: API Vulnerability Disclosures, Responsible Disclosure, Prompt Injection Research, Payload Sandboxing, OSINT

Security Advisories

ADV-2026-001: Prompt Injection Sandbox Escape (CRITICAL 9.8): RESOLVED

Identified a prompt injection flaw bypassing host executor sandbox on an AI inference platform.

ADV-2025-004: Cart Valuation Logic Bypass (HIGH 8.4): RESOLVED

Discovered state mutation flow allowing cart total manipulation via negative float injection in quick-commerce checkout API.

ADV-2025-002: OpenCV Frame Skip Bypass (MEDIUM 6.1): PATCHED

Bypassed face boundary state tracker via rapid frame injection crashing sleep detection algorithms.

Connect with Tanish Anand

Website: tanishanand.dev
GitHub: github.com/Tanish-Anand1
LinkedIn: linkedin.com/in/tanish-anand24
X / Twitter: x.com/sullaxive
Email: atanish920@gmail.com