Artificial intelligence enthusiastic#fullstack developer#PhD(AI)

2/2020 – 6/2021

1. scanned document recognition and digitalization with help of modern techniques like Tesseract, and computer vision.
Bulk scanned document processing and information extraction using Computer Vision and tesseract OCR a 1.5-year project (for Daimler at Stuttgart)
Main Tasks:
I have customized the computer Vision model to the recognized document instead of object recognition (That was purely my own innovation, and the idea worked really superb)
• Modifying prebuilt tesseract-OCR model to retrain and adapt with German language.
• Aligning the computer vision model and tesseract-OCR to the customer's specific document needs
• denoising, straightening, enhancing, and augmenting the customer's diverse set of documents
• meeting the 99% accuracy of information extraction of very complex scanned images.
• building and mapping database to show the extracted information at the front end.
• Building REST API to interact with web services

Computer Vision

11/2017 – 12/2020

1. Financial distress predictions with deep learning and Natural Language processing: `
Tasks:
• Collected thousands of parameters of bankruptcy and financial distress prediction of the USA-based big companies using Deep Learning Models.
• Aim was to predict financial distress for big firms beforehand using financial many different financial parameters

Deeplearning4j

11/2017 – 1/2020

Augmented Report Generation with Transformer-Based Language Deep Learning Model and Natural Language Processing
Tasks
• Handling with the 18 billion text tokens to automatically generate small business reports
• I have Used Transformer-based Neural network architecture is the most advanced one used by the open AI team.
• Model is able to write artificial business-related text which is grammatically correct and without any human involved.

Deeplearning4j, Maschinelles Lernen, Natural Language Processing