High-Accuracy Offline Handwritten OCR System

We are urgently looking for an experienced Machine Learning / Computer Vision developer to deliver a robust offline handwritten OCR system that supports both Hindi and English handwriting with high accuracy.

Project Requirements:

Must accurately read handwritten text (not printed).

Mandatory language support:

Hindi handwriting
English handwriting
The solution must work offline (no cloud APIs).

We are open to:

A custom-built OCR system, or
An already-developed OCR model you can provide.

Deliverables must include:

Full source code access
Model architecture + weights
Complete setup and usage documentation
Should work on standard image formats (JPG/PNG) and scanned pages.
Can assist with fine-tuning on our own dataset.

Required Expertise:

Strong experience with handwritten OCR
Deep learning (CNN/RNN/Transformers)
Tools/frameworks: PyTorch, TensorFlow/Keras, OpenCV, Tesseract (if customized)
Experience with Indic languages OCR (highly preferred)