Preethi
Preethi Narayanan · Cupertino
Engineer, builder, and recovering perfectionist. Things I'm making and what I'm learning along the way.
A short bio.
I'm currently SVP Engineering at Afero, a connected home platform serving enterprise customers like The Home Depot.
Earlier, I was one of five founding engineers who built Amazon Echo and launched Alexa — before voice AI was a product category. My career has been spent at the intersection of hardware, software, and AI: most recently, building AI agents, visual detection models, and edge AI systems on Jetson Thor.
I like to learn, build, and have honest conversations about hard problems — technical or otherwise.
Things I've built.
A catalogue for your wine collection. Take a pic, add notes — AI fills in the details from the label. Ask for recommendations, log what you tasted after the bottle's been opened.
wine.pnvlabs.workReact/TypeScript SPA on S3 + CloudFront, talking to Python 3.12 Lambdas behind API Gateway with Cognito auth, DynamoDB for storage, S3 for wine images, Gemini 2.5 Flash for vision/text AI, and SES for notifications — all wired together with AWS SAM.
Auth is invite-only via a Cognito admin-create flow with HMAC-signed approve links; per-user monthly AI quotas tracked in a separate DynamoDB table; admin dashboard gated on a server-side email allowlist.
When you're in the thick of work, you do — you rarely reflect. Later, when it matters, the details are gone. Capture quickly, tag with attributes like influence, polish with AI, search by what mattered.
reflect.pnvlabs.workFrontend: React (CRA) bundled to static assets, hosted on S3 + CloudFront, with Cognito for invite-only auth.
Backend: Node.js 24 Lambda behind API Gateway, DynamoDB for stories + usage, Secrets Manager for the Gemini API key, Gemini 2.5 Flash for AI polish.