I am a data scientist and machine learning specialist with expertise in deep learning, computer vision, and natural language processing. My academic background includes a Master's in Data Science and a Bachelor's in Computer Engineering.
Currently working as a Computer Vision Researcher at Indiana University School of Optometry, I focus on developing innovative solutions for infant visual development research using monocular depth estimation techniques.
Previously, I worked as a Data Scientist at Imerit Technologies, where I engineered robust ETL pipelines, developed interactive dashboards, and led computer vision annotation projects.
I'm passionate about solving complex problems using data-driven approaches and creating machine learning solutions that make a real-world impact.
Implementation of Mask R-CNN to detect suitable regions in video frames and seamlessly insert advertisements without interrupting playback.
View Details
Web application that predicts daily restaurant inventory needs using machine learning to reduce food wastage based on historical usage patterns.
View Details
Research investigating the effectiveness of monocular depth estimation techniques using deep learning approaches for studying infant visual development.
View Details
Multi-modal approach for music genre classification combining audio features from spectrograms and lyrical analysis with BERT.
View Details
AI-powered platform for collecting, evaluating, and prioritizing innovation ideas within organizations, using AI to automate scoring and enable data-driven decision making.
Live Demo
Web-based chatbot that answers questions as Lord Krishna by referencing the Bhagavad Gita, providing wisdom with proper citations to source texts.
View Details
Developed a model to predict a user's personality type (MBTI) based on their social media posts using bidirectional GRU networks.
View Details
TensorFlow project using bidirectional LSTM to generate lyrics to a song based on seed text with 56.69% accuracy.
View Details
Sentiment analysis model to identify positive and negative movie reviews using various neural network architectures including CNNs, single LSTMs, and multiple LSTMs.
View Details
Image classification model for distinguishing between cats and dogs using transfer learning with MobileNet architecture.
View Details
Used transfer learning with various models (MobileNet, VGG16, DenseNet) to detect pneumonia from chest X-ray images with 86.22% accuracy.
View Details
Applied various classification methods to predict rock categories (Igneous, Metamorphic, Sedimentary) based on 19 different rock features.
View Details
Explored different dimensionality reduction techniques (PCA, t-SNE, LLE, MDS) on rock image dataset and compared performance of clustering algorithms.
View Details
Used transfer learning for custom image classification, achieving 85.5% accuracy with MobileNetV2 compared to 63.7% with a model trained from scratch.
View Details
Full-stack website providing ratings, reviews, and movie information with user authentication for submitting personal reviews.
Live Demo
Web-based system for managing football team players with database integration and dashboard for visualizing player attributes and performance.
View Details
Created a data pipeline using Microsoft Azure services to collect, transform, and analyze comprehensive data from the Tokyo Olympics.
View DetailsIndiana University School of Optometry, Bloomington, IN
Indiana University, Bloomington, IN
Focused on deep learning, computer vision, and natural language processing applications.
Imerit Technologies, Remote
Developed strong foundations in software development, algorithms, and computer architecture.
Feel free to reach out to me for collaboration opportunities, project inquiries, or just to connect!
mmanav.3101@gmail.com
Bloomington, Indiana