Kshitij Kodati

A passionate Computer Science graduate with hands-on experience in data analytics, software development, and applied machine learning. Skilled in SQL, Python data manipulation (Pandas, NumPy), and marketing KPI reporting alongside building predictive models and responsive front-end tools. Proven ability to extract business insights from complex datasets and translate them into actionable dashboards and scalable technical solutions.

• SyncCheck: Real-Time Deepfake Detection Extension
  1. Developed a browser extension utilizing an Audio-Visual Fusion Model to detect deepfakes during live Zoom and Google Meet streams.
  2. Integrated MediaPipe facial landmark tracking with Mel Spectrogram speech analysis to identify desynchronization with [X]% precision.
  3. Optimized inference latency to deliver real-time security alerts within [X] milliseconds of stream playback.
• Multimodal Early Detection of Autism Spectrum Disorder (ASD)
  1. Implemented a multimodal diagnostic pipeline combining ResNet-50 for facial behavior, CNNs for drawing patterns, and MLPs for clinical questionnaires.
  2. Designed a weighted decision fusion mechanism that improved early diagnostic accuracy to [X]%.
  3. Deployed a responsive screening application using Streamlit and OpenCV for real-time assessment by healthcare providers.
• Retinal Disease Diagnostic System
  1. Built a Convolutional Neural Network (CNN) pipeline to classify fundus photography for diabetic retinopathy, glaucoma, and macular degeneration.
  2. Achieved [X]% validation accuracy across [N,000]+ retinal images using transfer learning and data augmentation.
• House Price Prediction Platform
  1. Evaluated regression models including XGBoost and Random Forest over housing feature sets, reducing Mean Absolute Error (MAE) by [X]%.
  2. Built an end-to-end data pipeline with feature engineering, exploratory data visualization, and an interactive web demo.

AWS Academy Graduate - AWS Academy Machine Learning Foundations