Solving real-world problems with scalable ML systems
Machine Learning | Data Science | Software Development
I am currently in my final year pursuing a B.Tech in Computer Science from Institute of Engineering and Management, Kolkata (GPA: 9.18). I enjoy building intelligent systems that blend deep learning, data engineering, and human-centric design.
From personalized recommender systems at WEBEL to gamified LMS platforms at IIT Ropar, my work spans across predictive modeling, backend systems, and research-driven design.
ML, Web Development, and Data Engineering Toolkit
Deep Learning • Anomaly Detection • Real-time Data
Deep Learning • Time Series
A resource-efficient stock forecasting tool using co-movement-based clustering and shared-layer-LSTM and Ticker-aware-LSTM with custom embeddings models to predict Nifty50 trends, enabling a single model to handle multiple stocks with high accuracy.
Anomaly Detection • Backend • Blockhain • Real-time
A platform that assigns safety scores to crypto wallets by combining supervised (XGBoost) and unsupervised (Auto-Encoder) ML models with graph-based anomaly detection. The system factors in transaction behavior, wallet age, and KYC status to identify scam-linked wallets and assess risk in real time.
RAG • Chatbot • Blockchain
A secure, localized Retrieval-Augmented Generation (RAG) chatbot that integrates LLMs with document-based Q&A capabilities. It supports on-device model inference and utilizes IPFS (InterPlanetary File System) for decentralized storage providing enhanced security, privacy, and accessibility even in offline or remote environments.
Recommender System Development: Built a personalized hybrid recommender using user-based collaborative filtering on over 20M real service records, queried via PostgreSQL.
Algorithm Engineering: Designed a multi-source similarity model combining demographics, historical usage, and service co-occurrence with rule-based eligibility filters to recommend from a pool of 313 government services.
Interactive Deployment: Deployed a Streamlit-based dashboard for real-time recommendations with adaptive feedback loop for continuous personalization.
Built Google Sheets LMS: Developed a learning management system using Apps Script, automating academic data workflows and improving efficiency by 95%.
Web Integration: Created a Flask-based dashboard that visualized LMS statistics from Sheets and automated email notifications.
Gamification: Added badge rewards using proven strategies to improve user engagement.
Specialized in Machine Learning and Artificial Intelligence with focus on practical applications and research.
Studied in Science stream with focus on Physics, Chemistry, Mathematics, and Computer Science.
I'm always open to discussing new opportunities and interesting projects.
© Soumyadeep Basak.