xtLytics LLC

Data Scientist

Jan 2019 — Jan 2020Noida, IN

Role Overview & Impact

Invoice OCR: Built GCP Vision + Tesseract + Faster-RCNN pipeline to extract and structure tabular content into CSV/XLSX via geometric bounding box heuristics.
CADtnet: Developed edge TFLite model (VGG-19, MobileNet, ResNet) for mobile breast cancer triage on ultrasound images, deployed in LMICs.

Projects & Initiatives at xtLytics LLC

Detailed technical models and systems built during this tenure

2 Projects

Details

CADtnet

Jan 2019 - Jan 2020

Developed an edge TFLite deep learning model for mobile breast cancer lesion segmentation and classification on ultrasound images (~6,500 scans dataset), deployed in Low-Middle Income Countries (Guadalajara, Mexico).

TFLite

VGG-19

MobileNet

ResNet

U-Net++

Android

Python

Details

TableOCR

Jan 2019 - Jan 2020

State-of-the-art Invoice & Supermarket Bill OCR using GCP Vision + Tesseract + Faster-RCNN to extract and structure loosely-tabular content into CSV/XLSX via geometric bounding box heuristics for 50+ US gas station clients.

GCP Vision API

Tesseract OCR

Faster-RCNN

OpenCV

Python