I’m a software engineer focused on backend development, AI, and machine learning, working primarily with Python and C++. I build scalable and reliable systems with an emphasis on clean architecture, performance, and practical application of AI technologies.
LLM and RAG
LLM-Powered Job Matching Agent
Takes a user's CV and portfolio and finds relevant jobs using a hybrid retrieval pipeline. Pulls from PDFs, portfolio data, and live web search. Runs locally using LLaMA via Ollama, so there are no API costs.
Multi-Source AI Knowledge Retrieval
RAG system that pulls from PDFs, scraped web pages, and academic sources. Has fallback logic for missing or low-quality sources so it stays reliable with messy, real-world inputs.
Predictive Modelling
Dynamic Pricing Model
PyTorch regression model that reduced pricing errors by 30%. Packaged with FastAPI and Docker so it runs as a live prediction service, not just a notebook.
Fraud Detection System
Random Forest classifier trained on imbalanced payment data. Used SMOTE to handle class imbalance, which is the main thing that breaks fraud models in practice.
Image Classification App
CNN trained on CIFAR-10, served via FastAPI and Docker. Send an image to the API, get a prediction back. Clean end-to-end pipeline.
Real-Time Lane Detection
C++ and OpenCV system that marks lane boundaries in live video streams. Built with autonomous driving use cases in mind.
APIs and Full-Stack
Health Information Exchange Platform
Patient mobile app and hospital web portal that share records securely during emergencies. Spring Boot backend with geolocation and facial recognition for identity verification across providers.
Office Management System
Full-stack HR and finance tool with real-time messaging over WebSockets. Built for internal teams who need visibility across departments without switching between tools.
Finance and Payments App
Mobile payments app with JWT auth, transaction scheduling, and a Spring Boot backend. Focused on security and reliable handling of financial operations.
Low-Level Engineering
Packet Sniffing Tool
Raw socket tool in C++ that captures and inspects TCP, UDP, and ICMP packets. Useful for network diagnostics and low-level protocol analysis.
File Upload Server
C++17 HTTP server that detects file types by magic bytes, categorises uploads, and parses CSV files. Written to handle varied and sometimes malformed input without crashing.
Get in touch
Open to full-time roles, contract work, and interesting collaborations.