At Nference (Anumana.ai), I built ECG image models that detect 52 abnormalities and 8 diseases, and designed an end-to-end workflow from ECG image capture through quality checks and model inference. I also built a two-stage LLM workflow to label ECG samples against clinical notes.
At Nference (Pramana.ai), I built imaging algorithms for whole-slide pathology scanners, including a graph-based stitching pipeline that generated 5M images at 99.5% accuracy. Inline image-quality checks reached 98% accuracy across 5M slides and reduced manual review from 10% to 2%.
Earlier, at Spectral Insights Pvt. Ltd. (acquired by Nference), I developed algorithms for nuclei segmentation and cancer-cell grading, image-focus optimization, and blood-cell classification. I’m an inventor on 9 US patents in imaging and computer vision.

