Marcin Iwiński

Senior Full-Stack AI Engineer at ThoughtWorks

Led QA and user-acceptance testing for a Rhino3D plugin aimed at autonomous vessel hull design; coordinated 6 external designers; reduced backlog by 56% within 8 weeks. Designed automated geometry test suites in Python and PyTest to validate boolean operations and meshing across 12 CAD models, cutting manual validation time by 72%. Implemented cross-platform packaging using MSBuild, Inno Setup, and Docker to produce Windows and macOS installers, increasing release frequency to bi-weekly. Validated API stability against RhinoCommon and Rhino7 with regression tests across 3 OS/VM configurations, achieving 95% pass rate. Prototyped interactive Grasshopper components in C# to accelerate path-planning, cutting geometry-generation time by 47%. Instrumented telemetry and error logging with Sentry; captured 98% field exceptions and enabling 34% faster triage. Automated CI pipelines with GitHub Actions to run unit tests, packaging, and integrations for each PR, reducing merge lead time by 41%.

Senior Full-Stack Engineer at Endava

Orchestrated integration of a CAD plugin backend with cloud APIs using Node.js and Express; served 200+ design exports per day and reduced manual handoffs by 83%. Implemented geometry processing microservices in .NET to precompute mesh LOD and metadata, shortening model load time by 58% for web previews. Refactored legacy CI to GitHub Actions and containerized builds with Docker, cutting build flakes by 92% and accelerating CI runtime by 48%. Optimized PostgreSQL schemas to store CAD metadata and telemetry, improving retrievals for 10k+ artefacts and reducing median query latency by 67%. Built RESTful APIs and WebSocket endpoints to stream design updates to React/TypeScript frontends, supporting real-time previews for 4 concurrent users. Reengineered data pipelines to preprocess geometry for web previews.

Senior Full-Stack Engineer at BJS S

Led design-automation web service for Rhino3D plugin; enabled batch processing of 5k+ models monthly; reduced manual prepb by 88%. Created PostgreSQL-backed job queues and lightweight schedulers; built NUnit-based test suites; increased developer confidence and regression catch rate by 58%. Configured CI with GitLab CI and containerized test runners using Docker, reducing environment drift by 92%. Trained junior engineers on Rhino Common, Brep operations, and computational geometry best practices; improved team throughput by 27%. Built conversion utilities to transform between internal CAD formats and STEP/OBJ, lowering client conversion costs by 41%. Monitored production telemetry dashboards to detect geometry processing bottlenecks; decreased average job time from 18 minutes to 7 minutes. Resolved cross-platform build issues for Windows to Linux tooling and automated installer scripts; achieved 98% success rate for client deployments.