"Bridging the gap between raw data and intelligent, explainable systems."
GRU, RAG, Multimodal APIs, Prompt Engineering, Vector Search, MMR, TensorFlow, Keras, NumPy, Pandas, Matplotlib
Dagster, Milvus, Docker, Google Gemini API, SQL
Python (Advanced), C++, TypeScript, Bash, Linux, Git/GitHub, FastAPI
A clinical decision-support tool for Obesity Risk Estimation using Explainable ML. We evaluated multiple models (Random Forest, XGBoost, LightGBM) on the UCI dataset, optimized memory usage, and built an intuitive Streamlit interface featuring SHAP visualizations for complete model transparency.
An AI-powered study companion acting as an interactive tutor. Leverages Google Gemini for real-time concept explanation and summarization. Built on React and Vite for zero-latency interactions with a distraction-free UI tailored for students.
A high-performance, multilingual RAG system designed to ingest, deduplicate, and search through 10,000+ engineering research papers. Features SHA-256 exact deduplication, semantic chunking, Milvus vector storage, and MMR re-ranking using BGE-M3 embeddings.
A character-level Recurrent Neural Network designed to generate novel, human-sounding names. Trained on a 5.6M+ row dataset using TensorFlow/Keras. Implements temperature scaling (T=1.5) for creativity and custom benchmarking for repetition testing.
Active Member & Workshop Participant
Beyond academics, I am highly involved in the Centrale Tech club. I actively participate in hands-on workshops to bridge the gap between theoretical knowledge and practical engineering skills.
DataCamp
DataCamp
Coursera
Currently open for new opportunities, collaborations, and hackathons.