Financial Document Intelligence System
Capgemini
RAG-based intelligence system for Master Facility Agreements and Sanction Letters
2x2 VectorDB architecture, Text-to-SQL hybrid reasoning, agentic orchestration
Enterprise, research, personal, and open-source work
Capgemini
RAG-based intelligence system for Master Facility Agreements and Sanction Letters
2x2 VectorDB architecture, Text-to-SQL hybrid reasoning, agentic orchestration
Capgemini
Internal Model Context Protocol server for secure LLM tool integration in Databricks workflows
MCP server tools, internal app deployment, governed tool access for LLM-driven workflows
Capgemini
Multi-agent system for pricing intelligence, competitor benchmarking, and AI-generated marketing content
Independent Personal Project
Multi-agent AI platform for investment research with retrieval, quantitative analysis, market intelligence, analytics, and response validation agents
Framework-agnostic agent architecture, unified MCP server, hybrid RAG, RAGAS and Langfuse evaluation; built with Claude Code, OpenCode, and Mimo Code
Launchpad.ai / Fellowship.ai
Agent using Gemini Pro Vision achieving 97% accuracy for price discovery across 100K+ SKUs
Fallback logic (VectorDB → SerpAPI), 50% reduction in manual lookup
Master's Dissertation
ML model on Microsoft Malware dataset (8.9M devices) with GDPR-compatible explainability
66.7% accuracy, 50% memory optimization, regulatory-aligned XAI
NLP & XGBoost-based Streamlit app to classify electrical product PDFs
K-Means + PCA for customer segmentation and targeted marketing
LSTM-based RNN to generate song lyrics from seed text
NLP pipeline classifying IMDb movie reviews as positive/negative using Bag of Words and TF-IDF feature extraction
89.9% accuracy with Logistic Regression and Naive Bayes
Deep neural network classifying clothing images into 10 categories on the Fashion MNIST dataset
Neural network model to predict loan repayment status using Lending Club data
Regression model with PCA dimensionality reduction to optimize manufacturing test bench speed
Movie rating prediction using logistic regression, decision trees, and random forests on the MovieLens dataset
Added documentation examples for f_regression() and silhouette_score()