Native C++ LLM benchmarking suite built exclusively for Apple Silicon Macs. Leverages Metal GPU acceleration via llama.cpp, real-time hardware telemetry, HuggingFace model browser, and a glassmorphic dark UI with live Chart.js visualizations. Zero Node.js, zero Electron.
Project Page ↗ GitHub ↗A local-first ReAct agentic AI assistant with SSE streaming, multi-turn tool calling using Qwen3-8B via llama.cpp, and a Perplexity-style Vanilla JS frontend. Serper.dev search integration with FastAPI backend.
GitHub ↗Local AI image generation pipeline combining FastAPI, llama.cpp LLM inference, and ComfyUI headless with JuggernautXL. Runs fully offline on Apple Silicon with zero cloud dependency.
GitHub ↗AI powered travel planning agent that generates smart multi-day itineraries using real-time weather and Google Maps data. Built with FastAPI and Google Gemini 2.5 Flash.
GitHub ↗Multi-stage AI pipeline for real-time meeting understanding using Faster-Whisper, pyannote diarization, Ollama LLMs, and ChromaDB-based Retrieval-Augmented Generation.
GitHub ↗RL-based quadruped locomotion built at DRDO. PPO agent trained in MuJoCo physics simulation using JAX and Stable Baselines3. From random flailing to a stable, controlled gait.
GitHub ↗Backend infrastructure built with FastAPI to serve locally hosted large language models via structured APIs.
GitHub ↗Multimodal AI agent supporting text and image input using locally deployed LLMs, designed for privacy-focused interaction via Telegram and WhatsApp.
GitHub ↗Computer vision-based AI proctoring system using face detection, gaze tracking, and behavioral monitoring to ensure examination integrity.
GitHub ↗
Built a reinforcement learning-based quadruped locomotion system from scratch at the Ministry of Defense's premier electronics research lab. Designed reward shaping functions, trained PPO agents in MuJoCo, and iterated on gait stability across uneven terrain using JAX and Stable Baselines3.
Led AI/ML workshops, organized hackathons, and mentored junior developers. Built a hands-on curriculum around practical ML and full-stack projects to grow technical depth within the chapter.
Served as an executive member, contributing to technical event planning, coding workshops, and community engagement initiatives across the campus.
Managed public relations, student outreach, and external communications for university events and campus initiatives.
Represented the chapter as a Google-recognized campus champion, driving developer engagement through technical talks, study jams, and peer mentoring. Coordinated cross-campus events and fostered a community of aspiring developers around Google technologies.
Breaking down LLMs, local AI setups, hardware builds, and robotics for developers. Fast-format. No fluff. Just the signal.
Behind-the-scenes of projects, dev setups, and the messy process of going from idea to deployed AI system.