I am an AI Systems Engineer & Full-Stack Developer based in Bengaluru, India, shipping production-grade AI-native products end-to-end—from multi-agent pipelines to real-time dashboards. I focus on the "last mile" of AI: bridging the gap between raw models and practical, hardened applications in legal tech, forensics, operations, and cybersecurity. Currently rebuilding in public, ensuring every featured project is fully deployed, open-source, and live.
- 🧠 AI Systems Engineering (tool-use, multi-agent pipelines, RAG)
- ⚛️ Full-Stack Development
- 📡 Real-time Observability & LLMOps (WebSockets, dashboards, metrics)
- 🔐 Cybersecurity & Ops Intelligence
- 🕵️ Privacy-first AI applications & Digital Forensics
- ⚖️ Legal tech (contract analysis, risk flagging)
- 🤝 Agentic workflows & Autonomous systems
- 🛡️ Production hardening (auth, sandboxing, rate limiting)
- 🌍 Building in public with live demos and documentation
Enterprise-grade RAG SaaS platform featuring multi-tenant document ingestion (PDF/DOCX), asynchronous BullMQ background workers, semantic vector search with Qdrant, and real-time streaming AI chat using Google Gemini. Built for production scalability with Redis caching and JWT security. Stack: Node.js · Gemini AI · Qdrant · Redis · BullMQ · JWT | A production-grade ReAct Agent powered by Llama 3.3 70B & Groq. Features Hybrid Search (Vector + BM25), real-time Web Search, technical system tools, and interactive chat-based training. Stack: Python · Groq · Llama 3.3 · Streamlit · BM25 |
Universal Model Context Protocol (MCP) server containing 10 typed system tools, a live React dashboard, and native Claude Chat integration with WebSocket streaming. Stack: Python · FastAPI · Pydantic · WebSockets · React · Claude | 💎 ONYXProduction-grade edge AI inference platform running fully in the browser with zero backend, local model execution, and a cyber-themed mission-control interface. Stack: HTML5 · CSS3 · JavaScript · WebGPU · ONNX Runtime |
Multi-agent AI research pipeline using specialized sequential agents (SCOUT, ANALYST, CRITIC, and SCRIBE) to crawl the web, extract findings, challenge logic, and compile structured reports. Stack: Python · LangChain · CrewAI · Tavily API · Gemini | ⚙️ GearFlowAutonomous, zero-trust bike service ecosystem orchestrated via n8n workflows. Integrates AI-driven dispatching, vision-based spare part verification, and a RAG pipeline for support. Stack: n8n · Python · OpenCV · RAG · PostgreSQL |
🕸️ LangraphAI Parliament Cloud: a multi-agent, real-time decision intelligence SaaS platform orchestrated using LangGraph to simulate collaborative debate and consensus building. Stack: Python · LangGraph · FastAPI · WebSockets · React | Autonomous local multi-agent system that eliminates documentation drift by automatically synchronizing code modifications with technical docs in real-time. Stack: Python · LangGraph · ChromaDB · Qwen 2.5 (7B) · Git Hooks |
Multi-role digital forensics lab containing 32 analysis modules (fingerprint, deepfake, etc.), a local Mistral chatbot, SQL case tracking, and automated ReportLab PDF generation. Stack: Python · Flask · OpenCV · Ollama (Mistral) · SQLite · ReportLab | AI Personal Memory Operating System featuring an immersive 3D spatial user interface, semantic constellations, and interactive exploration of personal data archives. Stack: HTML5 · JavaScript · Three.js · TensorFlow.js · WebGPU |
👁️ The WitnessReal-time non-dual intervention engine and deep system telemetry dashboard featuring instant telemetry updates and low-latency state synchronizations. Stack: Next.js · Node.js · WebSockets · TailwindCSS | Observability and monitoring platform designed to track LLM prompt executions, latency metrics, token consumption, and system costs to detect operational anomalies. Stack: FastAPI · Pydantic · InfluxDB · Grafana · WebSockets |
This is my 2025 rebuild after losing access to my previous GitHub account. Everything pinned above is what I've shipped since — every project is deployed, every README is real, every live demo is reachable. The graph will thicken from here; the work speaks for itself in the meantime.
⭐ If any of the projects above saved you time — or you want to talk MCP, agents, or AI-native product work — star a repo or drop me a line.