Engineering self-correcting RAG systems, Model Context Protocol servers, autonomous desktop copilots, and deterministic agent governance frameworks.
Specializing in AI-driven automation, data pipelines, and system orchestration. Experienced in integrating LLMs with low-latency workflow automation to eliminate operational bottlenecks and optimize high-volume processing.
Bachelor of Computer Science Engineering with AIML (2022–2026)
Techno India University
Stateful cyclic retrieval graphs with hybrid dense/sparse search and local neural cross-encoder re-ranking.
Hardware-aware model context protocol servers with real-time CPU, RAM, and VRAM telemetry integration.
High-throughput automated ingestion nodes, concurrent scrapers, and structured data serialization pipelines.
Decision-tree halting mechanisms and visual verification loops ensuring verifiable AI coding execution.
A showcase of self-correcting RAG architectures, hardware-aware protocol dispatchers, financial copilot platforms, and autonomous tooling.
An enterprise-grade, fully local, self-correcting agentic retrieval system. Built on stateful LangGraph cycles merging ChromaDB dense vectors with BM25 sparse lexical search, FlashRank neural cross-encoder re-ranking, and Moondream vision models for extracting embedded technical PDF diagrams.
Edge-native Model Context Protocol control plane and autonomous tool dispatcher. Built with a native Rust (Tauri v2) shell and Python FastMCP sidecar operating under a 100MB RAM overhead, bridging local LLMs with dynamic hardware telemetry (CPU, RAM, VRAM, I/O).
Autonomous financial AI workstation and desktop copilot. Combines a lightweight Tauri v2 Rust host with React/TypeScript and a headless FastAPI backend for real-time market modeling, background Core Cache Node telemetry, and multi-LLM engine routing.
Enterprise governance and evidence-driven skill engine for autonomous AI coding agents. Zero-dependency framework enforcing formal decision-tree halting, side-by-side DOM verification, and Test-Driven Defect Isolation to prevent assumption drift.
AI-powered smart agriculture architecture. Highly optimized, decoupled full-stack machine learning system (Next.js and FastAPI) computing optimal crop yields through microclimate soil telemetry, sensor analysis, and true probability mapping.
A reactive desktop media downloader powered by Electron and yt-dlp. Features an ultra-clean minimalist interface, native system integrations, automated CI/CD builds via GitHub Actions, and high-performance concurrent stream capture.
Available for AI system architecture, enterprise agent pipelines, and high-performance engineering collaborations.