About Me

I spent ten years learning why systems fail. I now do that for AI.

I'm Manvi Sharma — an applied AI engineer focused on making agentic and LLM systems reliable enough to ship inside regulated enterprises. For nearly a decade I built and validated mission-critical systems at Workday, AWS, IBM, and British Telecom, where the job was proving that complex software behaves correctly before it reaches production. I now bring that same reliability discipline to AI: the model is rarely the hard part — earning enough trust that a real organization will run the system is.

I've shipped across regulated domains — industrial & chemical manufacturing, enterprise HR & financials, healthcare, and telecommunications — each with its own auditors, failure costs, and definition of "safe." I hold a Master of Business Analytics (MBAN) from UBC Sauder, where I also served as a Graduate Teaching Assistant, and I collaborate on research applying AI to surgical decision-making.