PORTFOLIO
Hey, I'm Bharani Kumar. An enthusiastic AI full-stack developer who loves to build state-of-the-art web and mobile applications. I love turning concepts into scalable products using tech stack like React, FastAPI, Node.js, and artificial intelligence technologies. I always strive to keep my code clean and design great user experience along with learning new technologies like LLMs, LangChain, and MCP.
Currently a 2nd-year B.E. CSE student at Kingston Engineering College, Vellore. 55+ public repositories, 70+ AI projects, and a passion for real-world impact.
Building end-to-end AI-powered applications from autonomous agents to real-time screen analysis. Experience with 70+ AI projects including NOVA, IRIS, and SocialGuard-RL.
Crafting modern, responsive web interfaces with smooth animations and intuitive user experiences. From sleek dark-themed UIs to interactive 3D portfolios.
Scalable backend systems with robust API design, real-time data handling, and efficient database management for high-performance applications.
End-to-end AI solutions from data preprocessing to deployment. Specialized in LLM agents, computer vision, and intelligent automation systems.
Building responsive cross-platform mobile applications with smooth performance, intuitive interfaces, and seamless integration with backend services.
Streamlined deployment pipelines, containerized applications, and efficient development workflows for production-ready systems.
As a full-stack developer using modern ideas, simplicity design and universal visual identification tailored to dedicated and current market.
Autonomous AI system with voice, vision & multi-agent execution. Persistent, provider-agnostic AI runtime with Observe → Remember → Reason → Act → Verify loop.
Real-time screen AI assistant that follows your cursor to assist and teach you anything on your PC live. Hyper-powerful AI with contextual screen analysis.
Full-featured search engine in Rust. BM25 + PageRank ranking, BPE query language, Tor mode, 228+ unit tests, AI reranking, encrypted search, neural embeddings.
124M-parameter decoder-only LLM built from scratch. RoPE, RMSNorm, SwiGLU, GQA, KV Cache, BPE Tokenizer, full training pipeline. Research-grade quality.
Real-time object detection & video intelligence. YOLO + ByteTrack + FastAPI. Detects objects in images/videos/live webcam with persistent ID tracking.
Free public civic tech for Tamil Nadu land ownership. Patta, chitta, A-register, FMB sketches, interactive maps. Next.js 14 + FastAPI + Supabase. ₹0/month cost.
AI-powered fantasy cricket platform. 22,062 matches of data. Multi-model ensemble (XGBoost + LightGBM + CatBoost). Smart team optimizer with LLM explainability.
Internet Intelligence Platform. Search, browser execution, extraction, monitoring, memory, graph reasoning, cited answers. One programmable layer for web intelligence.
Decentralized P2P sharing platform. Files, folders, links, chat over encrypted WebRTC. Zero server storage, AES-256-GCM encryption, no accounts required.
Open infrastructure standard for AI agent networks. Agent registry, discovery, messaging (NATS), capability negotiation, reputation system, shared memory.
AI model scanning entire codebase for all kinds of bugs automatically. Comprehensive static and dynamic analysis pipeline for code quality assurance.
Photograph old circuit boards → AI reverse-engineers full modern rebuild blueprint. Maps every component and trace for hardware reconstruction.
Multimodal chatbot with voice, vision, chat & image generation. Conversational AI with multi-modal input/output capabilities.
AI misinformation detection & news credibility analysis. Classifies REAL/FAKE with confidence scores + explanations. Model-agnostic architecture.
Browser-based OSINT intelligence tool. AI-generated recon methodology, attack surface analysis, DNS/IP/breach lookups with auto target detection.
Late-night voice chats with Live2D AI companion. Real-time voice chat, emotion-driven expressions, lip sync, conversation memory via ChromaDB.
Real-time mobile voice changer with 14-stage DSP engine. Demon, Robot, Crushed Radio, Glitch presets for VoIP calls. Flutter + Kotlin native DSP.
System-wide real-time voice changer for Android/iOS. 10-layer DSP: gain boost, echo, crackle, distortion, bit crusher, pitch wobble, reverb, hard clipping.
AI-Powered Creator Command Center. GitHub stats, daily AI progress narratives, voice TODOs, hackathon mission control, context-aware AI copilot.
RAG-powered document intelligence with privacy-first design. Local processing, multi-format support, session isolation, entity masking, full audit trail.
Personal Operating System PWA. Local-first, offline-first. Unifies habits, tasks, workouts, study planning, hydration, nutrition, sleep, finance, CRM, calendar.
Autonomous CLI coding agent with soul. Textual TUI, 8 commands, Groq LLM, ChromaDB memory, NetworkX knowledge graph, 3 themes, AutoDebug, AI commits.
Persistent AI runtime architecture spec. 50+ documentation folders covering every subsystem from vision to disaster recovery. Provider-agnostic design.
AI-Powered Presentation Generator. Pure HTML + CSS + Vanilla JS — no frameworks, no build tools. Clean, minimal, production-ready slides.
Premium Notion-inspired Sudoku workspace. Pure HTML/CSS/JS. 9 game modes, smart hints, AI solver, daily challenges, full offline support.
Paste your code → get brutally roasted by AI → refactored into production-ready code. Fun + functional code review tool.
Dev Meme Generator with fresh/trending memes, custom text overlay, and one-click download. Built for developer humor and sharing.
Powerful privacy-first support assistant. Complete privacy, no data tracking, intelligent responses for user support scenarios.
Smart automation and vault management system. Secure storage with automated workflows and intelligent organization.
Magic and illusion-themed interactive project. Creative exploration of visual effects and interactive storytelling.
VANTA version 5 iteration with enhanced design system, improved animations, and refined user interface components.
Interactive 3D Book Portfolio. Vanilla HTML/CSS/JS with 3D page-flip book experience, chapters for bio, projects, skills, experience, certificates.
AI handwriting-style notes creator. Upload data, generate handwriting-style notes, print assignments. Educational productivity tool.
Real-time crowd density analysis using computer vision. Safety monitoring system for public spaces and events.
Automated exam correction via OCR and LLM evaluation. Intelligent grading system for educational institutions.
Smart scheduler with AI-generated summaries and action tracking. Automated meeting notes and task extraction.
AI travel itinerary builder from personal preferences. Smart trip planning with recommendations and budget optimization.
Swiggy-like food ordering app with full restaurant and cart system. Complete food delivery platform with real-time tracking.
Voice-based AI legal aid for underserved users. Accessible legal assistance through natural language voice interactions.
Human presence detection via WiFi signal patterns — no cameras required. Privacy-preserving occupancy sensing technology.
