Project Case
Nanobanana DL
Project built to generate images in batch with Nano Banana from lists of virtual gift ideas. The focus was productivity and outcome: automate sequential creation using the free Gemini API, operated via VS Code and CLI.
Challenge
Create a fast flow to generate many consistent images from prompts at zero cost, with sequence control and file organization. The project had to remain internal, prioritizing execution speed and predictable outputs over public UI.
Solution
Node automation consuming Gemini via Google AI Studio, with a batch prompt pipeline. The flow reads idea lists (virtual gifts) and generates images in sequence, with folder and naming conventions for traceability. Operations run in both VS Code and CLI to speed up iteration and execution.
My Role
Defined the flow strategy, wrote prompts and automation, integrated the Gemini API, and validated image quality. I ran tests, tuned parameters, and documented the operation for reuse in other projects.
Results and Impact
Delivered a working batch generator, reducing manual effort and production time. Using the free API kept costs at zero, and sequential generation ensured consistency and speed to feed another project.
Lessons Learned
Batch prompt engineering benefits from standardized templates and parameters, plus strict naming conventions to avoid rework. Combining VS Code and CLI made iteration faster and comparisons clearer.