Muhammad Usman

Software Engineer/Machine Learning Engineer

Arbisoft • Full time

Sep 2018 - Mar 2025 • 6 yrs 6 mos

Software Engineer / Machine Learning Engineer — Arbisoft Lahore, Pakistan | Oct 2018 – Mar 2025 (6+ years) Over six-plus years at Arbisoft, I worked across backend development and machine learning on production systems used by real clients. On the backend side, I designed server-side architecture and REST APIs using Django, SQL, and Celery — including work on the McKinsey Academy platform, a large-scale learning system where I optimized database performance and built the APIs connecting front-end and backend services. On the ML/AI side, I built and deployed AI chatbots (SMS, email, web) integrating LLMs and Twilio, including personalized bots that replicated specific personalities with voice support and vector-based context. I also built a Whisper-based speech-to-text pipeline and an AI-powered CV parsing tool, and trained/deployed ML models using TensorFlow, PyTorch, and Transformers via Docker and FastAPI on Azure. I also led FinOps-focused DevOps work to reduce cloud costs and automated end-to-end ML pipelines.