Muhammad U.

Architected and deployed scalable, AI-driven backend systems with FastAPI, incorporating advanced LLMs such as GPT-4 and Gemini Pro for core enterprise applications. Engineered multi-agent architectures and autonomous workflows that automated complex business processes. Designed production-grade automation pipelines using n8n for AI document generation, intelligent data extraction, and event processing. Developed vector search and semantic embedding pipelines to enhance internal data matching systems. Optimized the performance and latency of LLMs, minimizing API overhead and computational costs.

Led a technical team in the creation of dynamic AI agents utilizing LangGraph and n8n, facilitating integrations across corporate LLM services. Engineered and deployed machine learning models on Azure Machine Learning, creating robust infrastructures for ML training and inference. Integrated containerized ML models through REST APIs for high-availability consumption in web and mobile applications. Deployed Azure OpenAI (ChatGPT) solutions aligned with prompt engineering and security practices. Streamlined operations by orchestrating automated workflows via Azure Logic Apps and Azure Automation. Established monitoring solutions through Azure Monitor to ensure uptime for ML models and backend services. Developed corporate Q&A bots using Azure Cognitive Services and OpenAI APIs to reduce customer support ticket volume. Created executive analytics dashboards in Power BI to convert telemetry data into actionable KPIs.

Developed multi-threaded web scrapers with Scrapy and Selenium to extract unstructured breast cancer patient data from medical databases. Architected data preprocessing and feature engineering pipelines to prepare clinical datasets for advanced modeling. Evaluated classic machine learning classifiers, implementing optimized Support Vector Machines (SVM) on structured datasets. Advanced deep learning methods by applying Convolutional Neural Networks (CNNs) with pre-trained VGG16 weights to improve cancer classification accuracy.