Preet Dalal
B.Tech Information Technology (Honours in DevOps) · D. J. Sanghvi College of Engineering
I'm an IT student at D. J. Sanghvi College of Engineering, specializing in DevOps and Cloud Engineering, with backend development as my foundation and applied AI/ML as an additional strength.
I build backend systems in Java and Python, then containerize and deploy them, working through Docker, Kubernetes, and monitoring with Prometheus and Grafana along the way. Recent projects include a fraud detection platform running on Kubernetes with autoscaling, and a deep learning model for skin lesion classification that stays robust under adversarial attacks, both involved deploying and operating real ML systems in production, not just training them.
I'm currently deepening my Kubernetes and CI/CD skills, moving from single-service deployments toward more complete infrastructure setups.
Education
Dwarkadas J. Sanghvi College of Engineering
Expected 2028Projects
- ›Built a multi-channel fraud and phishing detection platform analyzing SMS, email, and URLs in real time, using TF-IDF classifiers and a Random Forest model for scoring.
- ›Deployed the FastAPI and React services on Kubernetes with a Horizontal Pod Autoscaler to absorb bursty inference traffic, and instrumented a /metrics endpoint scraped by Prometheus and visualized in Grafana.
- ›Designed a custom Residual Attention network (2.88M params) for skin lesion classification, trained with a curriculum adversarial schedule to resist adversarial perturbations.
- ›Built a multi-stage image defense pipeline (bit-depth reduction, blur, JPEG compression) that raised FGSM-attacked accuracy from 24% to 62.5%, plus a clinical decision support layer with automated PDF reporting.
- ›Built for Smart India Hackathon: a PyTorch U-Net segmentation model and EfficientNet-B0 classifier that extract building footprints, roads, and waterbodies from drone orthophotos.
- ›Implemented custom GIS vectorization (contour extraction, Shoelace area calculation) and deployed the pipeline across three services with model weights synced from Hugging Face Hub at runtime.
- ›Developing a Spring Boot REST API covering JWT authentication with role-based access, Redis caching, Flyway-managed migrations, and Testcontainers-based integration tests, containerized with Docker Compose.
Skills
Awards
UGEE Rank 53
Ranked 53 in the University Grants Entrance Examination
Get in Touch
Open to Backend, DevOps & MLOps Opportunities
Whether you have an internship opening, software engineering role, or want to discuss distributed backend systems and Kubernetes infrastructure, feel free to reach out directly through any of the channels below.