Open to opportunities

Shyamal Rana

@shyamalrana

Message

Computer Science graduate building deep learning and full-stack solutions.

India

Message

What I'm looking for

I’m looking for an entry-level role where I can build and deploy deep learning models for image segmentation and also ship full-stack features, with strong mentorship and clear opportunities to grow.

I’m a Computer Science & Engineering graduate from NIT Rourkela, with internship experience spanning deep learning, computer vision, and full-stack web development. I built and deployed a biomedical nucleus segmentation pipeline that reached 90.8% pixel-level accuracy on the 2018 Data Science Bowl dataset.

In my Machine Learning Internship, I trained a UNet with Squeeze-and-Excitation (SE) blocks and Atrous Spatial Pyramid Pooling (ASPP) for multi-scale feature extraction. By addressing class imbalance with a composite loss (Jaccard Distance + Dice Loss + Focal Loss), I raised F1 to 0.76 and pixel-level accuracy to 90.8% on the test set.

I also enjoy shipping end-to-end products: during my Software Development Internship, I built and shipped Angular + Spring Boot features and owned the full request cycle from UI to SQL queries. Alongside that, I’ve created projects like TaskForge (React/Node/Express/RBAC APIs) and a stock analysis dashboard with ARIMA-based forecasting.

Experience

Work history, roles, and key accomplishments

Software Development Intern

KW Solutions Private Limited

May 2024 - Jul 2024 (2 months)

Developed and shipped Angular and Spring Boot features for client-facing web applications, owning the request cycle from UI components to SQL queries. Wrote Selenium/Cucumber UI test suites to automate regression checks across three payment gateway integrations, reducing manual verification time per release cycle.

Angular Spring Boot REST API Development SQL Selenium UI Test Automation

Machine Learning Intern

National Institute of Technology, Rourkela

Jan 2024 - Apr 2024 (3 months)

Built a biomedical nucleus segmentation pipeline using a UNet with Squeeze-and-Excitation blocks and ASPP, improving multi-scale feature extraction. Implemented composite loss to address class imbalance, raising F1 to 0.76 and pixel-level accuracy to 90.8% on the test set.

Python UNet Squeeze And Excitation (SE)Atrous Spatial Pyramid Pooling (ASPP)Segmentation Composite Loss (Jaccard Dice Focal)U Net

Education

Degrees, certifications, and relevant coursework

National Institute of Technology, Rourkela

Bachelor of Technology, Computer Science & Engineering

2021 - 2025

Activities and societies: Relevant coursework: Data Structures & Algorithms; Machine Learning; Deep Learning & Artificial Intelligence; Computer Vision; Database Management Systems; Operating Systems; Computer Networks; Object-Oriented Programming.

B.Tech in Computer Science & Engineering at NIT Rourkela (2021–2025), with coursework spanning data structures, machine learning, deep learning, and computer vision.