Projects
My personal and professional projects demonstrating backend engineering expertise, scalable systems, and AI integration:
Transaction categorisation and financial coaching for Revolut
A four-stage pipeline built in 24 hours at HackDelft 2026 with Team YUBRE: a hybrid rules and sentence-embedding categoriser (75% accuracy, under 50 ms), life-stage personalisation, an Isolation Forest subscription engine, and a Claude Haiku money coach with one-tap action cards.
AI-powered restaurant food logistics platform
Forecasts ingredient demand and recommends how to cut food waste, with plain-English queries over the data. A microservices system built with FastAPI, RabbitMQ and KEDA autoscaling on Kubernetes, with LLM provider failover. Built for TU Delft's Software Architecture course.
Gesture-controlled video calling application
Developed a gesture recognition system integrated with real-time video communication. Showcases system design thinking—handling concurrent connections, low-latency processing, and reliable media streaming. Blends frontend and backend expertise with focus on performance.
Serverless media service with intelligent scaling
Engineered a production-grade media handling service focusing on reliability and cost optimization. Implemented efficient concurrent processing, automatic scaling, and intelligent resource allocation. Demonstrates core backend expertise in building scalable systems.
Building infrastructure for high-volume logging
Speaker at PyDelhiConf 2023 on designing log streaming systems. Deep dive into building reliable, scalable logging infrastructure—a foundational component of distributed systems. Covers architecture decisions, trade-offs, and production considerations.
AI-powered home management system for shared households
Designed and built an intelligent home management platform that automates inventory tracking, streamlines grocery ordering, and enables fair expense splitting. Demonstrates agentic AI capabilities in solving real household problems with practical impact.