At Indian Institute of Technology (IIT) Roorkee, I designed and built CloudBench, a platform comparing compute, storage, and networking across AWS, GCP, and Azure using Terraform and Python. I also architected a three-tier AWS infrastructure with EC2, RDS PostgreSQL, and S3/CloudFront.
I engineered CloudSnip, a Dockerized Flask URL shortener on EC2 with asynchronous Lambda logging and a GitHub Actions CI/CD pipeline. I’m conducting multi-cloud performance benchmarking research on latency, cold starts, database response times, storage throughput, and auto-scaling.
I implemented an MLOps pipeline for brain tumor segmentation with UNet++, using DVC, MLflow, Evidently, and DagsHub for remote storage, with deployment on Hugging Face Spaces. In a separate project, I fine-tuned a UNet++ model with TensorFlow/Keras and deployed a public Gradio inference app.
At Allegheny College, I conducted a comparative benchmarking study of text-to-image models and developed a standardized evaluation methodology using complex prompt engineering. My work and projects draw on cloud infrastructure, automation, and machine learning operations.
