Drone-inspection imagery pipeline: a legacy .NET worker re-implemented as Python 3.12 services on AWS (SQS, Lambda, ECS Fargate, S3) using OpenCV and libvips, verified against the legacy output. Measured on real inspection data: 16–20× faster tile generation on the same machine, ~10× faster per blade on AWS (17–31 s vs 281–737 s), ~10× less CPU time and task memory down from 30 GB to 8 GB. 800+ automated tests.

Event-driven .NET platform on AWS used by a global energy company: about 24 services and thousands of high-resolution images per inspection. ASP.NET Core REST APIs with multi-tenancy, role-based access and OAuth2/OIDC single sign-on; ECS Fargate workers for PDF reports and deep-zoom tiles; SQS queues and Lambda for archiving and retention; SQL Server, MongoDB, Elasticsearch and S3; React/TypeScript and Angular web apps; Docker and multi-environment CI/CD. Led by our architect for 6+ years.

.NET worker services on AWS that turn inspection data and high-resolution imagery into large PDF reports without blocking the web app. The ASP.NET Core API queues a job in SQS, an ECS Fargate worker renders the PDF and stores it in S3, and the user gets an email when it is ready. A scheduled Lambda and SQS-driven jobs archive and clean up binary assets according to retention rules.

Platform for searching and analyzing millions of inspection images on serverless AWS. My part: React features (virtualized thumbnail grid, bulk metadata editing, linked search filters, user and group admin, ML prediction overlays) and the serverless video pipeline: S3 and EventBridge start a Step Functions workflow, FFmpeg and MediaConvert produce animated previews and DASH streams, and a Lambda API signs every segment URL.