At Navikenz, I built Python and FastAPI services for GES Salesforce Automation that clean and enrich exhibitor data using LLMs and D&B. I replaced inconsistent free-form model responses with structured outputs that passed every UAT test and shipped to production.
I also replaced manual steps and standard Salesforce Flows with a Python and Apex pipeline using Salesforce Bulk API, and wrote deduplication logic to map enriched data to existing records. In my current GES Master Data Management work, I’m building validation and cleansing APIs and mapping the application data model to MDM pipelines on Reltio.
For a pneumonia detection project, I fine-tuned EfficientNet-B0 on chest X-rays and served the model through a FastAPI endpoint with a React frontend deployed on Azure. I also built a Microsoft Teams scheduling assistant backed by FastAPI, running quantized Phi-3 Mini locally.
