At SFM Technologies, I built a RAG system for telecom tariff audits using a fine-tuned Mistral 7B model. It reached 92% accuracy and reduced manual review time by about 40%, running fully on-premise so client data stayed within the company.
For my graduation project at Concordia University, I built Python pipelines using the Gerrit REST API and developed ReGerrit, an ML tool that predicts which rebases need CI re-runs. The work showed that dependent changes merged 2.3 times slower than standalone ones and was accepted at ESEM 2026.
I also built a Transformer model for tweet credibility and sentiment at RIADI Laboratory, and have developed projects in real-time factory monitoring, explainable multi-agent e-learning, AWS cost optimization, and text-to-ArchiMate generation. I hold an AWS Certified Solutions Architect – Associate certification and have published work accepted at ESEM 2026.
