Completed Projects

Built a rugged, offline-first GoPro backup and integrity verification system for riders and adventure travellers. It pairs a React/Vite dashboard for backup stats and AI travel journals with a native Python/Qt client that supports GUI and headless CLI workflows across laptops and Raspberry Pi devices.

Created a DAG-based task runner that executes dependent jobs with bounded parallelism and an optional coordinator-worker distributed execution layer over gRPC. The project is designed to showcase scheduler design, execution guarantees, observability, and distributed systems tradeoffs.

Developed an ML-powered therapy chat application with a Flask backend, Next.js frontend, and Ollama integration. Messages are saved instantly, model replies are generated asynchronously in the background, and each signed-in user keeps persistent chat history across mobile and laptop-friendly sessions.

Built a Raspberry Pi homelab registry service that tracks hosted apps, allocates ports from a fixed pool, and syncs service state with Docker labels and event streams. It acts as an operational control layer for keeping self-hosted services organized without polling-heavy overhead.

Designed a local browser dashboard for Raspberry Pi 5 that records interview question-answer pairs, transcribes them with Whisper, and cleans transcripts with Ollama before exporting structured text data. Audio stays in memory, which keeps storage lean and the workflow training-data focused.

Built a rugged, offline-first GoPro backup and integrity verification system for riders and adventure travellers. It pairs a React/Vite dashboard for backup stats and AI travel journals with a native Python/Qt client that supports GUI and headless CLI workflows across laptops and Raspberry Pi devices.

Created a DAG-based task runner that executes dependent jobs with bounded parallelism and an optional coordinator-worker distributed execution layer over gRPC. The project is designed to showcase scheduler design, execution guarantees, observability, and distributed systems tradeoffs.

Developed an ML-powered therapy chat application with a Flask backend, Next.js frontend, and Ollama integration. Messages are saved instantly, model replies are generated asynchronously in the background, and each signed-in user keeps persistent chat history across mobile and laptop-friendly sessions.

Built a Raspberry Pi homelab registry service that tracks hosted apps, allocates ports from a fixed pool, and syncs service state with Docker labels and event streams. It acts as an operational control layer for keeping self-hosted services organized without polling-heavy overhead.

Designed a local browser dashboard for Raspberry Pi 5 that records interview question-answer pairs, transcribes them with Whisper, and cleans transcripts with Ollama before exporting structured text data. Audio stays in memory, which keeps storage lean and the workflow training-data focused.

Developed a cross-organization desktop hiring platform for candidates, recruiters, and organizations, with role-based authentication and a MySQL-backed Python/PyQt6 application.

Built an unsupervised learning system to cluster emotions from speech audio using the RAVDESS dataset. The project extracts acoustic features, applies clustering, and explores a real-time emotion recognition workflow in Python.

Designed a wireless IoT sensor platform using ESP32 microcontrollers to capture and transmit real-time health and environmental data with modular PlatformIO-based C++ code.

Built an AI-powered execution engine for Git operations with Ollama integration to guide users through repo workflows and command generation from natural-language intent.

Coded a classic terminal Hangman game in C with hints, custom word lists, points-based rounds, and configurable settings for replayable learning and entertainment.

Built a terminal job scheduler in C with priority-queue and round-robin execution, live timers, keyboard navigation, and CRUD operations for task management on Windows.

Simulated a Unix-like file system in C with support for directory and file operations, metadata handling, and robust error paths to reinforce low-level systems concepts.

Developed a Raspberry Pi-hosted SaaS platform that lets users upload PDFs, read them in the browser, create notes, and maintain a TODO list from a lightweight self-hosted interface.

Created a Python-based music scheduler for Raspberry Pi that plays designated playlists by time of day, supports automation flows, and keeps CSV logs for reproducible routines.

Developed an intelligent farming system integrating IoT sensors, fog computing, and AI to assist farmers with crop disease detection, soil health monitoring, and irrigation management through a natural language-based interface.

Implemented an AI-powered resume classifier using BERT and logistic regression for automated job-role prediction from uploaded PDF resumes, wrapped in a PyQt desktop interface.

Built a Next.js starter application configured for Vercel deployment with TypeScript and modern font integration for fast full-stack prototyping.

Built a local-first, repo-native task tracker that stores task state inside the repository so history stays versioned with Git and works from any editor setup.