• 🔭 Currently Building: Voraxx — AI-powered VAPT platform with Shannon multi-agent engine + a dedicated Voraxx MCP Server for security tool orchestration
• 🆕 Recently Shipped: voraxx-mcp-server (CVE/Shodan/Nuclei tools over MCP), mcp-registry-finder (searches the official MCP registry from inside an agent session), and FrameSage (tool-calling agent for pandas DataFrames)
• 🧪 Latest Batch (Jul 2026): 10 new cross-domain tools — see 2026 Original Projects below; newest (Jul 16): ollama-audit, mcp-sentinel, and agent-tool-audit
• 🤖 AI/LLM Focus: Agentic AI systems, durable multi-agent pipelines (Temporal.io), RAG, and LLM-powered security automation
• 💡 Expertise: Full-stack development, AI agent architecture, security research, microservices, and cloud-native systems
• 👯 Collaborate With Me On: AI security tools, open-source agent frameworks, research papers, and hackathons
• 🤝 Help With: MCP server design, LangGraph orchestration, Mastra.ai workflows, backend optimization, and RAG pipelines
• 🌱 Currently Learning: Temporal.io durable workflow patterns, A2A (Agent-to-Agent) protocol, Google ADK multi-agent pipelines, and advanced LangGraph + Mastra.ai composition
• 💬 Ask Me About: AI-driven VAPT, agentic pipelines, MCP server design, multi-agent orchestration, and Shannon engine architecture
• 🎯 2026 Goals (H2): Ship Voraxx v1.0 + MCP Server (Q3), publish AI-assisted VAPT research write-up, open-source The-Ideal-Harness, and speak at a security/AI event
• ⚡ Fun Fact: Built a complete AI project prototype in 24 hours during a hackathon that got featured!

• Voraxx 🔒 Private · v1.0 in progress — AI-powered VAPT platform; automates recon, vulnerability scanning, and exploit chaining via LLM agents · TypeScript · Temporal.io · Claude Agent SDK
• Shannon AI Engine 🔬 Core R&D — Multi-agent reasoning engine inside Voraxx; orchestrates recon, enumeration, and exploitation workflows via LangGraph and MCP tools with durable Temporal.io workflows
• voraxx-mcp-server 🆕 — Minimal, dependency-free MCP server exposing CVE lookup (OSV.dev), Shodan host recon, and local Nuclei scan orchestration as agent tools · Python

• PRISM-Insfranest — AI backend-generation platform: NL prompts → YAML DSL → production-ready Django, Go Fiber, or Rails projects with Docker, tests, and docs; benchmarked across 2,664 generations · React · TypeScript · Flask · Groq
• The-Ideal-Harness — Claude Code-native agent harness with MCP-portable enforcement and safety tooling, 170+ tests · TypeScript · MCP
• FrameSage 🆕 — Tool-calling analysis agent for pandas DataFrames; answers plain-English questions and generates EDA reports with an explicit anti-hallucination boundary, no API key required · Python
• Task-Tokenizer ⭐ — Web3 gig-economy platform with on-chain job listings and wallet-based freelancer matching · Solidity · Hardhat · Next.js
• mcp-registry-finder 🆕 — Zero-dependency MCP server for searching the official Model Context Protocol registry from inside an agent session · JavaScript
• Quorix — Local-first reference app for agentic UI patterns: streaming replies, inline tool-call visualization, generative UI · React · TypeScript
• GenAI-Platform — Generative AI platform experiments and integrations
• Mentoro — Gamified AI learning companion · AI · Full-Stack
• NomadAI — Voice-first, multilingual AI travel companion for Delhi; Whisper + Groq LLM + Telegram · Python · FastAPI

📂 See all projects at github.com/bharat3645

Ten new, independent tools shipped across July 2026 — each one deliberately combines two or more of my fields (security, AI agents, dev tooling, NLP, ML, SRE, AI infra) to solve a specific real-world problem, with a real tested implementation behind every README, not just a pitch.

• dep-graveyard — Composite dependency-risk scorer: blends CVE severity (OSV.dev), package staleness, and license risk into one A–F grade with concrete remediation advice, instead of just another raw CVE list · Python · Security + Dev Tooling + optional AI narrative
• citeguard — Fully offline citation-claim verifier for research papers: flags citations that don't actually support the sentence next to them, using transparent TF-IDF similarity — no black-box API, nothing leaves your machine · Python · NLP + Academic Integrity
• a11y-agent — Context-aware accessibility scanner for the semantic smells axe-core and eslint-plugin-jsx-a11y miss: generic alt text, vague link text, color-only meaning, heading-hierarchy skips · Python · Web A11y + Static Analysis + AI
• moodmesh — Private, ethics-first engineering-team health detector: turns git commit timing and PR review sentiment into rising-risk trend flags for managers, not individual scores · Python · Dev Analytics + NLP
• biasscope — Lightweight ML fairness report card: demographic parity, equal opportunity, and disparate-impact (four-fifths rule) computed transparently in pandas, graded A–F in plain English · Python · ML + Compliance
• shiftsense — Turns a raw incident chat timeline into a structured blameless-postmortem draft: phase detection, time-to-mitigate/resolve stats, and deduplicated action items · Python · SRE + AI
• ai-origin-trace — Transparent, fully offline stylometric AI-vs-human authorship signal for text and code — explainable per-feature breakdown instead of a black-box percentage, built with explicit responsible-use framing against false accusations · Python · Forensics + AI Ethics
• ollama-audit 🆕 — Offline storage-hygiene auditor for a local Ollama model store: finds broken models (missing blobs), stale models, and orphaned reclaimable blobs by reading manifests/ and blobs/ directly — zero dependencies, zero network calls · Node.js · AI Infra + Dev Tooling
• mcp-sentinel 🆕 — Offline risk scanner for MCP client configs: grades each configured server A–F on inline secrets, @latest pins, shell indirection, and typosquat-like package names — zero network calls, zero dependencies · Python · Security + AI Agents
• agent-tool-audit 🆕 — Offline static analyzer for AI agent tool definitions (MCP tools/list or OpenAI function specs): flags prompt-injection-suspect descriptions, broad scope claims, and schema hygiene issues, graded A–F · Python · Security + AI Agents

I write about what I'm building and learning:

• 🛡️ AI-powered VAPT — multi-agent security automation with LangGraph
• 🔗 MCP server design — extending LLMs with custom security tools
• 🤖 Multi-agent system design and orchestration patterns
• ⚡ Durable AI workflows with Temporal.io

📝 Notes and posts in BLOG.md

• 🏅 AWS Cloud Practitioner
• 🏅 Google Cloud Associate Cloud Engineer (in progress)
• 📖 Advanced Machine Learning Specialization
• 📖 Kubernetes Administration
• 📖 LLM & Prompt Engineering Mastery

📜 Full list in CERTIFICATIONS.md

🟢 Open to: AI agent projects · Security research · Hackathons · Consulting

I'm always interested in:

• 🤝 Collaborating on AI agents, security tooling, and open-source projects
• 🔬 Co-authoring research on LLM-assisted VAPT and agentic systems
• 💼 Freelance & consulting: AI integrations, RAG pipelines, MCP server design
• 🎓 Sharing knowledge — reach out for mentorship, talks, or pair programming
• 🚀 Building impactful solutions that ship fast

If you find my projects helpful, consider supporting me:

Made with ❤️ by Bharat Singh Parihar

Building Voraxx — AI-Powered VAPT · Shipping 2026