VIJAY SARADHI NAVULURI
Summary
MSc in Artificial Intelligence graduate with hands-on experience in designing and implementing generative deep learning models, including GANs for text-to-image synthesis. Proficient in Python, PyTorch, and end-to-end ML workflow development. Seeking an AI Engineer role to contribute to innovative AI-driven solutions and scalable machine learning systems.
Overview
2
2
years of professional experience
Work History
Oracle Finance Consultant
IBM
Pune
08.2017 - 08.2019
- Gained experience in data extraction, transformation, and system optimization—skills transferable to data preprocessing and pipeline development in ML projects
- Designed and implemented Oracle Financial modules including General Ledger, Accounts Payable, and Accounts Receivable to streamline financial operations.
- Developed and maintained complex PL/SQL scripts and reports to support financial data analysis and regulatory compliance.
Education
MSc - Artificial Intelligence
Royal Holloway, University of London
10-2021
BTech - Electronics & Communication Engineering
Siddaganga Institute of Technology
01.2017
Skills
- Python and SQL
- Version control with Git
- Containerization using Docker
- Data analysis with Jupyter
- Integrated development with VS Code
- Deep learning frameworks: PyTorch, Keras, TensorFlow
- Machine learning libraries: Scikit-learn, Pandas, NumPy, Matplotlib, SciPy
- Modelling techniques: Regression, Classification, Decision Trees, Ensemble Methods
- Neural networks: CNN, RNN, Transformers, GANs, Attention Mechanisms
Projects
Text-to-Image Synthesis Using StackGAN, cGAN, and WGAN | Python, PyTorch, Matplotlib, NumPy
• Developed and compared three GAN variants (StackGAN, cGAN, WGAN) for generating bird and flower images from textual descriptions using CUB-200 and Oxford-102 datasets
• Implemented attention mechanisms in StackGAN to improve fine-grained details, achieving 22% improvement in Inception Score compared to baseline cGAN
• Fine-tuned WGAN with gradient penalty to stabilize training and reduce mode collapse, improving FID scores by 18% over standard GAN training
• Preprocessed and augmented 30K+ image-text pairs, implemented custom data loaders, and visualized training progress with loss curves and sample generations
Timeline
Oracle Finance Consultant
IBM
08.2017 - 08.2019
MSc - Artificial Intelligence
Royal Holloway, University of London
BTech - Electronics & Communication Engineering
Siddaganga Institute of Technology