Umesh Uddar

Data Scientist

PROJECTS

1. Image Classification Model Development [Python, TensorFlow, Keras]

   Developed an image classification model in Google Colab to classify images as "Happy" or "Sad." Achieved 93.5% accuracy using a convolutional neural network (CNN).

2. Forecasting Economic Indicators and FTSE 100 [Python, ARIMA, Exponential Smoothing]

   Forecasted UK economic indicators and FTSE 100 index using ARIMA and exponential smoothing models. Provided robust predictions aiding informed decision-making.

3. Web Content Analysis: Extracting Insights from Website Text [Python, LangChain, OpenAI]

   Utilized NLP techniques with LangChain and OpenAI to process and analyze web content. Demonstrated proficiency in text chunking, vectorization, and similarity search.

4. Kidney Disease Classification [Python, TensorFlow, MLflow, DVC]

   Developed deep learning models to classify kidney disease, leveraging MLflow and DVC for experiment tracking and data management. Integrated CI/CD pipelines for deployment.

5. AI-Driven Solution for Managing Duplicate Images in Digital Archives [Python, TensorFlow]

   This project explores the development and implementation of an AI-driven solution to detect and manage duplicate images within large-scale digital archives. Focused on the digitized collection of British Royal Naval ships at the National Museum of the Royal Navy (NMRN), it leverages advanced AI models—Perceptual Hashing (pHash), Inception V3, and Xception.

6. Real Estate Analysis Project

   Analyzed campaign performance across cities and channels, identifying key metrics and developing data-driven strategies for optimizing future campaigns.