Dr. Azad Abad

• Developed & demo an End‑to‑End Training Data Platform (TDP) in Azure Cloud Service to centralize all the documents & their corresponding labels in Azure Data‑lake
• Design and implement document ingestion pipeline using Azure Data Factory
• Design and implement document classification pipeline to train, deploy, and monitor machine learning models using Azure ML studio

• Developed & demoed an End‑to‑End pipeline prototype for extracting various data fields from Leased Contracts by down‑streaming & fine‑tuning BERT transformer language model (MLM) in question answering task. (Python, Pytorch, Flask, java script, HTML, MLFlow)
• Collaborated with all the team members to build an End‑to‑End pipeline for classifying real‑estate assets and extracting data fields from PDF and Excel sheets. (SVM, Python, Flask, Airflow, Docker, AWS, S3, EC2)
• Designed, implemented, and evaluated the Machine learning-based signature detection application for legal contracts using advanced object detection solutions. (e.g. Fast R‑CNN & Yolo v3)
• Trained and deployed a machine learning model in production for extracting name entities from documents using bidirectional LSTM (BI‑LSTM) with a bi-directional Conditional Random Field (BI‑CRF) layer. (Keras, Python)
• Created and presented models for Blueprint image classification using hybrid approaches. (Python, sklearn)

CogNet Project

• Building an Intelligent System of Insights and Action for 5G Network Management
• Applying machine learning to predict the traffic of5G network based on social media content (e.g. Twitter)

Dante a Teatro e Smart Milano Project

• Design and implemented an internal crowdsourcing platform for collecting and generating training data
• Implemented an End‑to‑End question classification pipeline for a closed domain task (food and dining habits in the Middle Ages) in Italian

IBM jeopardy

• The project was a research collaboration between IBM and the University of Trento for a deep understanding of question classification tasks and by encoding syntactic similarities of the questions using Tree Kernels

• Implemented a core application for banking automation solutions.(C#, SQL, Windows, Visual Studio)