Blog
Practical write-ups from RSA Team's work on AI, speech, computer vision, and production ML systems.
A practical summary of adapting Whisper Large v3 for Serbian audio pipelines, focusing on dataset preparation, evaluation, and production tradeoffs.
2024-04-01 · 9 min read
Designing a detection-plus-OCR pipeline is less about model choice alone and more about ROI quality, post-processing, and error handling in real documents.
2023-09-01 · 10 min read
