Devu Raju Kurup

Hey, I'm Devu

I enjoy coding, researching ideas, and building things that work well and feel thoughtful. Whether it's full-stack apps, ML models, or distributed systems, I like figuring things out and putting them together.

About Me

Hi, I'm Devu Raju Kurup. I'm currently pursuing an MSc in Artificial Intelligence and Data Engineering at University College London.

I have around three years of experience as a software engineer, where I worked on a client project and internal products, building features, fixing bugs, and collaborating across teams. Most of my work was in full-stack development using React JS and Ruby on Rails.

These days, I'm more interested in artificial intelligence and deep learning, especially how models are built into real-world applications and scaled efficiently through distributed pipelines. With what I'm learning in AI and data engineering, I'm excited to bring my background into new ideas and hands-on projects. I'm open to collaborations, research roles, or anything that helps me grow and build better systems.

I'm currently based in London, still exploring the city one neighborhood at a time. Whether it's walking with music in my ears or discovering cozy cafes, every corner tells a story. Outside of tech, I love cooking (often recreating my mother's recipes or experimenting with continental dishes), drawing, crafting, and reading. Capturing small moments like a sunset, a street light, or a face in the crowd is something I hold close.

That's a bit about me. Thanks for stopping by :)

Projects

Distributed Toxicity Classification

A scalable distributed system for classifying GitHub comments using a transformer model. Includes FastAPI UI, Spark processing, and Prometheus-Grafana monitoring.

FastAPIApache SparkAnsibleTerraformPython

Weakly Supervised Segmentation

Investigated WSSS improvements via class-adaptive thresholds and noise injection. Used GradCAM, CRF, FCN-ResNet, DeepLabV3, and Optuna tuning.

PyTorchSemantic SegmentationCRFOptuna

Protein Structure Analysis System

Deployed a Spark-YARN pipeline on a 5-node cluster for distributed 3D protein analysis. Used Merizo search, MinIO storage, and Grafana monitoring.

Apache SparkTerraformAnsiblePrometheusMinIO

Scribble – Article Publishing Platform

A content publishing platform with article management, scheduling, version tracking, and category-based organization. Built using React JS and Ruby on Rails.

React JSRuby on RailsPostgreSQL

Quiz Management Application

An app for creating quizzes, exporting results, and managing reports. Built with React JS and Ruby on Rails with clean UI and user analytics.

React JSRuby on RailsPostgreSQL

Video Summarization for IoT Surveillance

Created a tool to summarize surveillance videos using deep CNNs and hierarchical weighted fusion for high-quality, frame-relevant summaries.

Deep LearningPythonOpenCVComputer Vision

Let's Connect

I'm always open to opportunities, collaborations, or just a friendly hello. Feel free to reach out if something sparks your interest!

devukurup199918@gmail.com