Optimizing Neo4j queries that take so much time affecting the customer.Automation of adding Network Elements to sites using CSVs and APIs instead of UI Manual Network Addition.

-Provided a better solution for forced alignment using Montreal Forced Aligner instead of Aeneas Forced Aligner with results in English alignment around 140 times better. Error decreased from 591 wrongly aligned sentences (Aeneas result) to 4 wrongly aligned sentences (Montreal result).-Created Arabic acoustic model to use with Montreal Forced Aligner in aligning Arabic videos. Where we used 80 hours of audio to train and validate the acoustic model with lots of experimentation to get best results. Montreal Arabic aligner decrease number of wrongly aligned sentences to fifth compared to Aeneas Arabic aligner.-Trained our In-House Tashkeel model using Project Shakkala with a corpus 1 million words and run training for 99 epochs to get a WER (word error rate) of 1.86% and DER (Diacritics error rate) of 0.61% [Test data is 10% of the corpus size].-Helped in creating a simple Django web app using for Shakkala deployed on a Linux EC2 instance.-Helped in creating a simple Django web app using Django for IBTool deployed on windows using technologies like Celery(python task queueing) and Redis(message broker) for task queueing and flower to monitor the tasks.

Alignment project where our role were to align books to our Mathematical content. We were aligning k-12 grades for lots and lots of publishers for American, British, Indian and lots of other countries Mathematics curriculum.Linear Algebra Skillmap project: I created a skillmap for Linear Algebra were I break the topic into subtopics and then lessons and align our In-House lessons to these divided lessons reviewing its scoop and creating a lesson description for.Linear Algebra Lessons Reviewing and I worked that project as overtime after I worked as a python developer in Nagwa.Reviewing lessons and creating question distractors was our main tasks as a Mathematics content developer.