Ali Fadel

Ali Fadel

Software Development Engineer

Amazon

Biography

I am a software development engineer at Amazon. I am interested in machine learning research, specially, natural language processing (NLP) for Arabic language and source code related problems. In parallel, I am working on multiple small open-source projects to learn more about software engineering. Also, I am a Youtuber (A small one), I am teaching problem solving, basic machine learning and other topics on my channel, YAGs.

Interests
  • Natural Language Processing
  • Software Engineering
  • Problem Solving
  • Content Creation
Education
  • BSc in Computer Science, 2019

    Jordan University of Science and Technology

Experience

 
 
 
 
 
Software Development Engineer
Oct 2019 – Present Amman, Jordan
  • Developed multiple data pipelines that handle terabytes of data to provide daily training/inferencing ready builds.
  • Worked on developing Arabic and RTL supported systems either in the engineering part or by developing machine learning based solutions to do language identification and machine translation.
  • Participated in the design and the development of multiple search systems/experiments that support production traffic for multiple marketplaces.
 
 
 
 
 
Research Assistant
May 2019 – Present Irbid, Jordan
  • Worked on natural language processing (NLP) tasks such as text classification like Semantic Text Similarity (STS), token labeling like Arabic Text Diacritization (ADT) and sequence-to-sequence problems like neural machine translation (NMT).
  • Participated in machine learning competitions such as NSURL Semantic Question Similarity in Arabic (2nd/10), WANLP MADAR task (4th/19), SemEval ComVE task (3rd/17).
  • Organized multiple machine learning competitions like AI-SOCO at FIRE and ArEnMulti30K at WAT.
  • Published multiple research papers in various conferences.
 
 
 
 
 
Machine Learning Engineer Intern
Samsung Electronics
Feb 2019 – May 2019 Amman, Jordan
  • I worked on text classification problems using machine and deep learning techniques like TF-IDF, SVMs, RNNs, CNNs, and Transformers to do dialects identification that supports multi-dialect translation system.
  • The newly developed dialects identification system increased accuracy by 3% by proposing a new model architecture for Arabic language using RNNs and word embeddings.
  • Evaluated different word embedding techniques such as Word2Vec and FastText, and visualize them using t-SNE dimensionality reduction algorithm to find the best embeddings to use with Arabic words.
  • At the end of the internship, I initiated the noise-cleaning project that aims to automatically clean the noises from Bixby audio segments and classify the potential ones to be used as training data.
 
 
 
 
 
Freelance Trainer
Jul 2017 – Feb 2019 Remote
  • Recorded videos for three different courses that explain how to use Ruby on Rails framework to build real-life applications.
  • The first course was about the basics of Ruby on Rails and how to build a simple Content Management System (CMS) application.
  • Next, I developed a course about using Ruby on Rails framework to build a forum like HsoubIO (Similar to StackOverflow).
  • The last course was about scaling Ruby on Rails framework to build big projects like Twitter.
  • The courses are available through Hsoub Academy.

Accomplish­ments

Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning
See certificate
Deep Learning Spepcialization
See certificate
Machine Learning
See certificate
Systematic Program Design
See certificate
CS50: Introduction to Computer Science
See certificate

Projects

Shakkelha

Shakkelha

Research project to power Arabic NLP by building a system for automatic diacritization of Arabic text using deep learning techniques.

KONTESTS

KONTESTS

Web crawler brings the programming contests from many online judges and schedule them in one place.

codeforces2pdf

codeforces2pdf

Simple and easy to use tool to extract CodeForces contests and problems into PDF files in a readable and user friendly format.

Ruby3arabi

Ruby3arabi

The first Arabic website that explains Ruby programming language and provides online interpreter to allow users to run their codes while learning.

Publications

Quickly discover relevant content by filtering publications.