Devarya Lakhanpal

Workflow Automation > Building internal tooling and decision-support systems for telecom infrastructure project management, reducing manual coordination overhead across multi-site deployments. Developing full-stack features across the client delivery pipeline using React, Node, Python, improving real-time visibility into project status and resource allocation. Integrating AI-assisted workflows to surface anomalies and automate routine reporting, cutting analyst review time by 16%

CIP'6 > Selected from a global applicant pool to lead a cohort through Stanford's CS106A curriculum. Mentored 25+ students across the world, focusing on debugging, computational thinking and building projects.

Project ISF > Worked on a service delivery automation platform using RPA, streamlining multi-step workflows and reducing manual coordination across operational teams. Built a modular navigation interface in ReactJS and Node.js with RAG-powered contextual search across service documentation, enabling dynamic component rendering based on user role and context. Architected real-time frontend-backend sync via REST APIs and WebSockets, enabling live decision-support updates within the platform UI.

> Project Purity Built and trained a CNN on a self-collected, hand-labelled dataset for brand purity detection, achieving 96% accuracy and deployed into production for a global Fortune 500 CPG client. Led a team as engineer on Project Purity, scoping the ML pipeline from data collection through model training, API wrapping, and cloud deployment. Integrated computer vision inference into RESTful APIs and deployed on AWS (Ubuntu) using Nginx and Flask, enabling real-time product inspection at scale.

Asteroid Grand Challenge > Discovered a previously uncatalogued asteroid, 2016 JQ20 (P10vKns), now registered with the International Astronomical Union, as part of NASA's Asteroid Grand Challenge program. Analysed deep-space telescope imagery using image processing and pattern recognition techniques to differentiate asteroid candidates from background stars across thousands of image frames.

Intensive 180 hour course, offered in collaboration with KI-Campus and Stiftverband. Taught by Prof. Jan Peters.

Microdegree offered in collaboration with KI-Campus & Stiftverband. Taught by Dr. rer. nat. Michael Schwarzenberger. Relevant courses include: Machine Learning in Production, Deep Learning for Sequential Process Data & Processinformatics

Bachelor of Technology, majored in Computer Science and Engineering, specialized in Artificial Intelligence and Machine Learning