Roopesh Reddy

About Roopesh Reddy

Deloitte empowers my work designing advanced AI solutions using TensorFlow, PyTorch, and transformer models like BERT and GPT. My focus includes optimizing GANs for tasks such as image synthesis and domain-specific NLP applications, alongside leveraging retrieval-augmented generation techniques to enhance LLM factuality. By collaborating with teams, we deliver scalable AI innovations tailored to real-world needs. With a robust foundation in data engineering, I build ETL pipelines using PySpark and Spark SQL, ensuring high-quality data processing and transformations. My academic journey from JNTUH to the University of Missouri-Kansas City complements my technical expertise, driving innovative AI and data workflows. Passionate about advancing AI-driven solutions, I aim to bridge cutting-edge research and enterprise applications.

Experience

1. Senior Generative Ai Developer Lead

   Johnson & Johnson

   May 2024 — Present · Chicago, IL, US

   I design and implement AI solutions using TensorFlow and PyTorch, specializing in GANs (DCGAN, CycleGAN, StyleGAN) for image synthesis, style transfer, and data augmentation. I optimize models for convergence and stability, apply custom loss functions, and fine-tune transformer models like BERT and GPT for domain-specific NLP tasks. I\'ve also worked with retrieval-augmented generation (RAG) techniques to improve the factual accuracy of large language models.My data engineering experience includes building ETL pipelines with PySpark, using Spark SQL for data processing, and ensuring data quality with Pandas and NumPy. I\'ve developed encryption scripts using hashing algorithms, automated data workflows, and designed feature engineering strategies to support model training and analytics. I also build CI/CD pipelines to streamline deployment and version control.On the cloud side, I deploy models using AWS SageMaker and AWS Lambda for scalable, real-time AI applications. I’ve implemented prompt tuning for task adaptability and collaborated with teams to create industry-specific AI solutions. With a strong focus on performance, I apply model optimization techniques like quantization, pruning, and distillation for efficient deployment across cloud and edge environments.

Education

• University of Missouri-Kansas City

  Master's degree, Computer Science