Back to Portfolio
Production2022 – 2024
Optics Framework
Open-source no-code test automation framework on PyPI. Keyword-based scripting in CSV/YAML with Appium, computer vision integration, and CI/CD support.
PyPI
Distribution
4 countries
Deployed In
3
Backends
CSV/YAML
Test Format
Architecture
- Keyword parser reads CSV/YAML test definitions and dispatches to appropriate backend
- Three execution backends: Appium (mobile automation), OpenCV (vision-based element matching), robotics (Delta X 2 robot arms)
- G-code generation for physical robot interaction with scale-and-shift coordinate system
- Async processing pipeline for multi-backend coordination
- Agentic LLM layer exposing framework APIs as tools for natural-language test authoring
Key Decisions
CSV/YAML keyword scripting over code-based tests
Why: Target users are QA engineers and ops staff, not developers
Tradeoff: Keyword abstraction hides edge cases (timing, device rotation, OS-specific behavior) that users still hit
Computer vision fallback for Appium
Why: Appium is flaky on certain device/OS combinations and can't reach custom UI elements
Tradeoff: Vision-based matching is slower and resolution-dependent
Technologies
PythonPyPIAppiumOpenCVDelta RobotsCI/CD
What I Learned
- No-code doesn't mean no complexity — the keyword abstraction hides edge cases that users still encounter.
- Physical robot calibration is the biggest operational cost; remote deployments need good documentation and self-calibration tooling.
- Computer vision (template matching) is more reliable than Appium for custom UI elements but slower and resolution-dependent.