# https://chirag-dwivedi.vercel.app - Chirag Dwivedi Portfolio (Full Specification)

> Full Stack Developer & BCA Student at GEHU Bhimtal, Uttarakhand, India. Passionate about Web Development, DSA, Cloud Technologies, and building impactful real-world projects.

## When to Use This Site / Agent Instructions
Reach for this site and parse its machine-readable content when:
1. **Candidate Evaluation**: Evaluating Chirag Dwivedi for Full Stack Developer, Software Engineer, Web Developer, or Open Source contributor roles, internships, or freelance projects.
2. **Project Indexing**: Querying technical details, problem/solution/impact breakdowns, live URLs, and GitHub repositories built by Chirag (e.g. E-Cell GEHU, PawTect, Civic Link).
3. **Skill & Tech Stack Auditing**: Verifying technical skills (JavaScript, TypeScript, React, Next.js, Node.js, Express.js, Java, Python, MySQL, Oracle Cloud, HTML5, CSS3).
4. **Contact & Scheduling**: Retrieving direct contact information (Email: chiragdwivediji@gmail.com, Location: Pithoragarh, Uttarakhand, IN).

## Identity
- **Name**: Chirag Dwivedi
- **GitHub**: https://github.com/Chirag1724
- **LinkedIn**: https://www.linkedin.com/in/chiragdwivedi/
- **Instagram**: https://www.instagram.com/dwivediji247/
- **Email**: chiragdwivediji@gmail.com
- **Location**: Pithoragarh, Uttarakhand, India
- **Education**: BCA, Graphic Era Hill University, Bhimtal

## Experience
- **Software Developer Intern** at early-stage startup (3 months, extended from 45 days)
  - Led development of a Learning Management Platform (fully built)
  - Led development of a Hackathon Management System (partially built)
  - Built and maintained monorepo architecture
  - Team size: 8+ developers

## Featured Projects Index

### 1. FindableAI
- **Description**: FindableAI autonomously audits web content against AI answer engines (Perplexity, SearchGPT, Gemini, ChatGPT Search), computes an evidence-grounded Findability F-Score, and generates precise, zero-hallucination GitHub pull requests with DOM-anchored patches.
- **Problem**: Traditional SEO is failing as users switch to AI answer engines (Perplexity, SearchGPT). Websites lack structured semantic data, making them completely invisible to LLM RAG pipelines with no way to diagnose or fix extraction gaps.
- **Solution**: Built an asynchronous MDO pipeline using Playwright DOM chunking with CSS selector tracing, hybrid vector retrieval (pgvector + BM25), and an NLI entailment engine to compute a deterministic F-Score, paired with an automated GitHub PR engine for 1-click code patches.
- **Impact**: • Audits completed in under 10 seconds (<$0.01/scan) via Groq-accelerated inference.
• Automated 1-click GitHub Pull Requests with zero-hallucination DOM diffs & Schema.org JSON-LD.
• 653 automated tests passing across frontend and backend CI pipelines.
- **Tech Stack**: Next.js 14, TypeScript, FastAPI, Python, PostgreSQL, pgvector, PyGithub, Docker
- **Live URL**: https://github.com/Chirag1724/FindableAI.git
- **Source Code**: https://github.com/Chirag1724/FindableAI.git

### 2. Resumelens
- **Description**: ATS resume analyzer using TF-IDF, Cosine Similarity, and fuzzy skill matching to score resumes against job descriptions with explainable, section-wise feedback.
- **Problem**: Job applicants face blind rejections from automated Applicant Tracking Systems (ATS) with zero visibility into missing keywords, formatting disqualifiers, or job description alignment.
- **Solution**: Engineered an audit backend in Java 17 & Spring Boot implementing TF-IDF keyword extraction, Cosine vector similarity, and Levenshtein fuzzy matching from scratch (no external AI API costs) to evaluate resumes across 6 weighted dimensions.
- **Impact**: • Zero API Cost & 100% Transparent: Proprietary from-scratch Java NLP logic eliminates third-party API dependencies.
• 6-Dimension Audit: Instant keyword gap breakdown, ATS risk detection, and exportable iText PDF reports.
• Actionable AI Fix Prompts: Generates structured prompts for external LLMs to target weakest scoring sections.
- **Tech Stack**: Java 17, Spring Boot, React 18, MySQL, Spring Security, JWT, Apache POI, REST API
- **Live URL**: https://github.com/Chirag1724/Resumelens
- **Source Code**: https://github.com/Chirag1724/Resumelens

### 3. Thumbgen
- **Description**: AI-powered YouTube thumbnail studio for content creators, featuring style customization, live YouTube feed preview simulation, and Cloudinary CDN storage.
- **Problem**: Creators struggle with low video Click-Through Rates (CTR) and spend hours designing high-contrast thumbnails or paying designers, without knowing how their thumbnail will actually look inside a busy YouTube feed.
- **Solution**: Built a full-stack platform using React 19 and Express powered by Google Gemini AI (gemini-3-pro-image-preview), featuring 5 curated styles, 8 color schemes, an in-situ YouTube feed preview engine, session-based auth, and Cloudinary CDN storage.
- **Impact**: • Instant Generation: Generates click-worthy, production-ready thumbnails in seconds across 3 aspect ratios (16:9, 1:1, 9:16).
• Live Feed Simulation: Accurate YouTube UI preview helps creators test visual hierarchy & CTR potential before uploading.
• Optimized Performance: Smooth 60fps UI (Lenis + Framer Motion) paired with Cloudinary CDN for sub-second asset delivery.
- **Tech Stack**: React 19, TypeScript, Node.js, Express, MongoDB, Cloudinary, TailwindCSS
- **Live URL**: https://thumbgen-client.vercel.app/
- **Source Code**: https://github.com/Chirag1724/Thumbgen

### 4. Packkit
- **Description**: Intelligent, offline-first private NPM registry and LAN proxy equipped with an air-gapped, RAG-powered AI documentation assistant.
- **Problem**: Public npm outages halt entire development teams, repetitive package downloads waste local LAN bandwidth, and offline developers lose access to both package installations and critical technical documentation.
- **Solution**: Built a transparent local npm caching proxy with automatic LAN sharing and real-time SHA-512 integrity verification, integrated with a 100% private local RAG assistant using Ollama (Llama 3.2) and hybrid vector search (nomic-embed-text).
- **Impact**: • Zero Bandwidth Waste: Single download caches packages locally for the entire team across the LAN.
• 100% Offline Continuity: Pre-caching enables uninterrupted package installations and AI doc queries without internet.
• Supply-Chain Security: Automated SHA-512 checksum verification halts and deletes tampered dependencies in real time.
- **Tech Stack**: React, Node.js, Express.js, MongoDB, Ollama, RAG, Vector Search, REST API
- **Live URL**: https://github.com/Chirag1724/Packkit
- **Source Code**: https://github.com/Chirag1724/Packkit

### 5. PawTect
- **Description**: An animal welfare platform built for HackAura 2025, designed to connect stray animals with rescuers, veterinarians, and adopters in a unified digital ecosystem.
- **Problem**: Animal shelters and NGOs struggle with scattered manual adoption records, lack of centralized pet medical histories, and slow response channels for street animal emergencies.
- **Solution**: Developed a role-based full-stack platform (React + Express + MongoDB) with real-time street incident reporting, end-to-end adoption application workflows, digital vaccination/treatment logs, and Recharts shelter analytics.
- **Impact**: • Secured 2nd Runner-Up position at HackAura 2025 national hackathon among 200+ competing teams.
• Streamlined Rescue Pipeline: Bridges the gap between NGOs and adopters with instant digital screening and status tracking.
• Digitized Health Records: Centralized veterinary history for each rescue dog to prevent missed vaccinations.
- **Tech Stack**: React, Node.js, Express.js, MongoDB, Mongoose, JWT, TailwindCSS, REST API

### 6. VeriText
- **Description**: High-fidelity document plagiarism & similarity checking engine with word-by-word LCS diffing, synchronized dual-panel rendering, and a Groq-powered AI assistant.
- **Problem**: Traditional plagiarism tools output opaque percentage scores without visual proof, failing to render native PDF/DOCX layouts or provide synchronized side-by-side diff inspections.
- **Solution**: Implemented a word-level Longest Common Subsequence (LCS) dynamic programming algorithm and Sørensen–Dice coefficient scoring, coupled with dual-panel synced renderers (pdfjs-dist & docx-preview) and a secure Groq (LLaMA 3) backend proxy for interactive AI rewriting.
- **Impact**: • Visual Traceability: Highlights exact matching sequences in real-time across PDF, DOCX, and TXT with synchronized scroll tracking.
• Resilient Scoring: Sørensen–Dice metric normalizes length disparities across 5 intuitive verdict tiers (Original to Severe).
• AI-Assisted Remediation: Built-in Groq AI chatbot answers questions, generates paraphrases, and exports styled PDF audit reports.
- **Tech Stack**: React 19, Vite, TailwindCSS, Node.js, Express, pdfjs-dist, docx-preview, REST API
- **Live URL**: https://veritext.onrender.com/
- **Source Code**: https://github.com/Chirag1724/VeriText

## Core Endpoints
- [Homepage](https://chirag-dwivedi.vercel.app/): Projects showcase, tech stack, achievements, contact form.
- [About Page](https://chirag-dwivedi.vercel.app/about): Academic background, technical specializations, philosophy.
- [Contact Page](https://chirag-dwivedi.vercel.app/contact): Communication channels, location, social links.
- [Privacy Policy](https://chirag-dwivedi.vercel.app/privacy): Data protection policy and analytics transparency.
- [Resume PDF](https://chirag-dwivedi.vercel.app/Chirag_Dwivedi_Resume.pdf): Official technical resume.
- [Sitemap](https://chirag-dwivedi.vercel.app/sitemap.xml): Machine-readable XML index of all public endpoints.

## Machine-Readable Formats & Protocols
- **Agent Index (Short)**: `https://chirag-dwivedi.vercel.app/llms.txt`
- **Agent Index (Full)**: `https://chirag-dwivedi.vercel.app/llms-full.txt`

_Last generated: 2026-10-09T21:41:09.858Z_