Personal Projects
Here are some of the projects I have worked on.
PerceptoAI
Developed an intelligent voice assistant leveraging a Retrieval-Augmented Generation (RAG) pipeline for real-time, context-aware conversational AI, integrating dynamic location, weather, date/time and web search services.
Web-Based Framework for Generalizable Brain Tumor Segmentation
Built an in-house medical imaging platform, featuring advanced AI segmentation and LLM-based reporting capabilities through a model designed for generalizability across various brain tumor populations.
Multilingual Named Entity Recognition for Switzerland
Developed a multilingual NER NLP system using mBERT and XLM-RoBERTa transformers, evaluated on the PAN-X dataset. Compared monolingual and multilingual fine-tuning strategies across German, French, Italian, and English.
FindMe - Face Detection & Recognition
Developed a face detection & recognition system using eigenvector-based feature extraction and SVM, trained on the ORL dataset for accurate real-time identification across varied conditions.
Advanced Computer Vision Suite
Developed a full-featured desktop application from scratch, integrating advanced image processing and computer vision techniques—from basic noise manipulation to complex image segmentation.
Home Security Voice-code Access System
Developed a voice-activated desktop security app using DSP and machine learning, enabling secure access via spoken passcodes and voiceprint verification with real-time recognition and spectrogram visualization.
Fourier Transform Image Processing Mixer
Built a desktop app for interactive 2D signal mixing using Fourier Transform, enabling real-time image reconstruction by adjusting magnitude and phase frequency components.
Hospital Admin Prosthetics Website
Built an admin dashboard for prosthetics management, enhancing user experience with intuitive interfaces and efficient database CRUD operations.
EEG-Based Eye-Open State BCI System
Designed and implemented an EEG-based system with actuator control, incorporating signal preprocessing, feature extraction, and a Random Forest classifier for real-time response to open vs closed eye states.
