👋 As part of the engineering team, I had the opportunity to help bring Security Questionnaire AI to life. What excited me most about this project was solving for messy real-world input. Questionnaires don’t come in one clean format, they come as 200 row Excel sheets, PDFs, random Word tables, or those dreaded vendor portals. Making the AI flexible enough to parse and respond in all those environments was a big challenge. Another hard part: accuracy at scale. It’s not enough to just generate a “plausible” answer; every response had to be grounded in a company’s actual policies and systems. We designed the system so the AI only answers from verified sources (your policies, audit evidence, security controls) and nothing else. Honestly, seeing engineers and security teams get hours back each week because they’re not stuck answering “what’s your backup frequency?” for the 27th time has been super rewarding.🚀