I’m Divij Kapur, an engineer working across applied AI, backend systems, and product delivery. I turn ambiguous requirements into software that is useful, inspectable, and ready to operate.
SELECTED PROJECTS
Three projects that show how I approach product, infrastructure, and applied AI.
A human-in-the-loop financial complaint operations platform combining retrieval, policy evidence, and operator judgment.
I built ComplaintOps to learn what it takes to move a RAG idea beyond a chatbot. It ingests public CFPB data, retrieves historical cases and operating policy separately, produces cited recommendations, and keeps the final decision with a human reviewer.
An end-to-end phishing classification system built around reliable validation, reproducible training, experiment tracking, and API serving.
A modular training and serving pipeline with automated validation, transformation, model artifacts, monitoring, and a containerized delivery path.
A GPU-accelerated scientific pipeline that turns massive 3D tissue volumes into queryable marker co-localization evidence.
A reproducible and resumable Nextflow pipeline for analyzing spatial co-location of biological markers on NYU’s HPC environment.
EXPERIENCE
Product engineering, applied AI, and research—each with a different definition of “done.”
Capgemini Americas · New York
Engineered an LLM-powered proof of concept that transformed SRS documents into structured epics, user stories, and resource-aware sprint plans prepared for Jira integration.
U-NEXT Learning · Bangalore
Built APIs, real-time communication features, client SDKs, and video-processing workflows for remote proctoring and assessment products. Improved API response times by 40% and reduced video delivery from days to near-instant availability.
NYU VIDA · New York
Developed a reproducible GPU-accelerated pipeline for analyzing biological marker co-localization in 3D tissue images, with modular and resumable stages built for NYU’s HPC environment.
NYU Tandon · New York
Led graduate recitations in C programming and embedded systems, helping students build and debug STM32F429I projects with PlatformIO and Mbed OS.
ABOUT ME
I work across software engineering and applied machine learning, especially when a problem crosses boundaries: model and infrastructure, research and product, technical tradeoffs and user needs.
The details that make a prototype dependable matter to me—data quality, failure modes, authentication, evaluation, deployment, and interfaces that help people make good decisions.
EDUCATION
New York University, Tandon School of Engineering
Manipal University Jaipur
TOOLKIT
HAVE A HARD ENGINEERING PROBLEM?
I’m open to software, forward deployed, and AI engineering roles.
divij.kapur@nyu.edu ↗ Back to top