My name is Saketh Polavarapu, and I am pursuing an M.S. in Artificial Intelligence and Business Analytics at the University of South Florida. I’m passionate about applying AI and machine learning to solve real-world challenges. With hands-on experience in data science, automation, and cloud computing, I’ve built solutions that improve efficiency, accuracy, and decision-making.
I enjoy projects that bridge the gap between theory and practical impact, from predictive modeling to intelligent automation. My interests include deep learning, federated learning, and scalable cloud architectures, and I aim to contribute approaches that make technology more accessible and reliable.
Converts lip movements to speech via an AV-HuBERT pipeline: mouth ROI extraction, lip-reading, and Text-to-Speech with SpeechT5. Streamlit UI for upload and playback.
Tech: PyTorch · transformers · fairseq · OpenCV · dlib · Streamlit
Privacy-preserving image classification simulating multiple hospital clients with a central server using Flower. Fine-tuned ResNet50 and integrated Differential Privacy (tracked ε). Containerized runs for reproducibility.
Tech: PyTorch · Flower · DP-SGD · Docker
Email: sakethpolavarapu@usf.edu · LinkedIn: saketh-polavarapu · GitHub: Sakeh-Polavarapu
Audioless Video → Speech (AV-HuBERT)
Federated Pneumonia Detection (Chest X-rays)