Portfolio

Home / Portfolio

AI Healthcare System Content AI Healthcare System
Content

AI Healthcare System

AI Healthcare System: AI-Powered Medical Insights

A production-grade healthcare platform that combines Machine Learning for disease prediction with Generative AI for intelligent medical assistance. Built with a modern microservices architecture, the system provides preliminary health assessments for multiple conditions while maintaining strict data privacy.

Tech Stack

  • AI/ML: XGBoost, Gemini Pro (GenAI), Gemini Vision, FAISS (Vector DB)
  • Backend: FastAPI, Python, JWT Auth
  • Frontend: Streamlit
  • Infrastructure: Docker, Render (Backend), Streamlit Cloud (Frontend)

Key Features

  • Multi-Disease Prediction: Specialized models for Diabetes, Heart Disease, Liver Disease, Kidney Disease, and Lung Cancer.
  • RAG-Based Medical Assistant: Uses Gemini Pro and FAISS to provide context-aware responses grounded in medical data, minimizing hallucinations.
  • Smart Lab Report Analyzer: Utilizes Gemini Vision to extract and analyze data from PDF/Image lab reports automatically.
  • Secure Architecture: Microservices design with tenant isolation and secure session management.

View Project

Employee Attrition and Performance Analysis Content Employee Attrition and Performance Analysis
Content

Employee Attrition and Performance Analysis

Employee Attrition and Performance Analysis

A data-driven project analyzing employee turnover to identify driving factors and improve retention. It utilizes Powe BI for dynamic dashboards and actionable insights.

Tech Stack

  • Tools: Power BI, Excel
  • Techniques: Data Visualization, Trend Analysis, Dashboarding

Key Features

  • Attrition Factors: Identifies key variables contributing to employee turnover.
  • Dynamic Dashboards: Real-time visualization of workforce metrics.
  • Strategic Insights: monitoring trends to optimize productivity and retention.
  • Interactive Reports: Drill-down capabilities for detailed analysis.

View Project

Martin Ford Books Martin Ford
Movie Recommendation System Content Movie Recommendation System
Content

Movie Recommendation System

Movie Recommendation System Based on SBERT & FAISS

A sophisticated recommendation engine leveraging SBERT (MPNet) and FAISS for high-performance vector search. The system balances relevance with diversity using Maximal Marginal Relevance (MMR) and enhances user engagement through multi-factor re-ranking (Director, Franchise, Quality).

Tech Stack

  • Language: Python, PySpark
  • AI/ML: SBERT (all-mpnet-base-v2), FAISS (Facebook AI Similarity Search)
  • Backend: FastAPI (Async)
  • Frontend: Streamlit (Video Backgrounds)
  • Infrastructure: Render (Backend), Streamlit Cloud (Frontend), Docker

Key Features

  • Advanced Vector Search: 768-dimensional embeddings using MPNet for deep semantic understanding.
  • Blazing Fast Retrieval: Sub-100ms nearest neighbor lookups using FAISS.
  • Smart Re-ranking: Multi-factor logic (Director, Franchise, Era, Quality) to refine recommendations.
  • MMR Diversity: Balances relevance with diversity to prevent repetitive suggestions.
  • Rich Media UI: Premium dark-mode interface with trailers and poster integration via TMDB.

View Project

PRABC Content PRABC
Content

PRABC

Predicting the Risk of Attaining Breast Cancer

This project focuses on estimating the probability of developing breast cancer using advanced data insights and machine learning. By analyzing clinical data, the tool helps women understand their risk levels and the need for preventive care. It features a cross-platform informatic application (Mobile, Windows, MacOS) to make these insights accessible.

Tech Stack

  • Languages: Python, Java, Kotlin
  • Web Framework: Flask
  • Frontend: HTML, CSS, D3.js
  • Tools: Excel for data handling

Key Features

  • Risk Prediction: Uses a Random Forest Classifier trained on BRCA data to estimate cancer risk.
  • Cross-Platform Access: Informatic applications available for Mobile and Desktop OS.
  • Data Insights: Provides visual data representation using D3.js.
  • Performance: Model fine-tuned using precision and recall metrics for high reliability.

View Project