Sarika R

At Fiserv, I work as an AI/ML Engineer focused on Large Language Models and Generative AI solutions for retail and financial data use cases. I design and fine-tune GPT- and LLaMA-based models using PyTorch, TensorFlow, and Hugging Face to generate synthetic customer behavior data and support advanced analytics. My role involves building scalable training and inference pipelines on AWS (EC2, SageMaker, S3), developing REST APIs using FastAPI/Flask, and integrating models into enterprise systems. I also manage CI/CD workflows, model versioning, and automated retraining using MLflow, Docker, and GitLab CI/CD, while conducting rigorous model evaluation to ensure performance, coherence, and scalability in production.

At GNR IT Solutions, I contributed as an AI/ML Engineer developing predictive and generative machine learning solutions for logistics and supply chain domains. I built and fine-tuned Transformer and GAN-based models for demand forecasting and synthetic time-series generation using PyTorch, TensorFlow, and Hugging Face. I designed robust data preprocessing and augmentation pipelines with Pandas and NumPy to handle noisy real-world datasets, and deployed models on AWS and Google Vertex AI for scalable inference. My work also included API integration using Flask, model monitoring, continuous training pipelines, and collaboration with frontend teams to enable real-time forecasting and visualization.

Worked on an image-processing ML project focused on detecting aging patterns using Python. Performed data cleaning and preprocessing with Pandas/NumPy, trained and evaluated models with scikit-learn, and visualized results using Matplotlib. Gained hands-on experience in model experimentation, feature extraction, and dataset preparation.