Abd elrahman Hassanein
Work Experiences
Software Engineer
Research and Development Engineer
- Fine-tuned state of the art computer vision architectures for Object detection, Instance Segmentation and Semantic Segmentation tasks on AWS.
- Deployed state of the art Computer vision models for Object detection, Multi Object Tracking, Instance and Semantic Segmentation, and Monocular Depth Estimation on AWS.
- Used cloud services like AWS Batch and Lambda to automate visualizing results of deployed deep learning Inference endpoints.
- Developed, deployed and maintained a web app used internally by Meti’s employees to manage tasks, attendance, quality checks and time tracking. Used python, django, AWS, github and github actions.
Junior Machine Learning Engineer
- Contributed Exploratory Data Analysis notebooks that highlighted important aspects of the nature of the Archylise floor plan layouts data.
- Contributed Visualization notebooks that helped create visualizations of floor plan layouts and annotations.
Latest Projects
Finetuning,Serving Llama 3
In this repo I experiment with finetuning llama3 8b model on synthetic dataset created using gpt4-o. Followed by a full deployment to both synth streaming endpoint with langserve and async endpoint with FastAPI. Full article on the whole Process here.
HUB
Is an LMS platform, aims to help Educators, Teachers and Teaching service providers set up their platform, get up and running quickly with an LMS template.
AI Chat Companion
AI Chat Companion is part of HUB platform, it has information on the current article you are reading or course you are taking and is ready to help tackle any questions you have!
Content Recommendation
Similarity based recommendation feature, part of HUB platform to recommend content (articles, blogs, courses) based on semantic similarity.
ML-MLOPS Selected Projects
ML-Training-Pipeline
This porject illustrates a machine learning training pipeline. The main focus is the MLOPS Practices like Workflow Orchestration, CICD, Experiment tracking, Model versioning.
Named Entity Recognition API
An API to extract named entities (English) from text using bert-base-NER. Implemented with Flask.
Yolov5 Realtime and Async Deployment with AWS Sagemaker
Notebooks and code to Deploy Yolov5 to SageMaker realtime and async endpoints.
Backend-Dev Selected Projects
Jobs API
A simple RESTful CRUD API Implemented in MongoDB with Node.js+Express. Used docker to build API Container, and K8s to deploy and manage the containerized API.
PokiAPI-Graphql
A graphql API that uses PokeAPI as a datasource, implemented with Node.js and Apollo.
Frontend-Dev Selected Projects
GameShareApp
GameShare is a website that helps users rent their video games. Tools i used to build it include React, React-router, Vite, Bootstrap, Material UI and Firebase.this web app is inspired by the design here. you can visit it at: game-share.netlify.app
ML-DataScience Selected Projects
U.S. Patent Phrase to Phrase Matching - kaggle
Notebooks created for the kaggle PTP matching competition. EDA, Siamese LSTM network, 'Bert-for-patents' using Hugging face and Keras, Sentence-Transformers with 'AI-Growth-Lab/PatentSBERTa'.
Realtime Tweets Sentiment Analysis Using Spark Mlib and Spark Structured Streaming
A text sentiment analysis pipeline implementation using Pyspark to classify tweets polarity (positive/negative), then applying the Pipeline to streaming tweets from twitter API using spark’s structured streaming, and streams the output to parquet files.
Mask detection Training and deployment on AWS
Fine tuned the Faster RCNN Resnet50 model from object detection model zoo on the Mask Detection Dataset then Deployed the trained model to AWS SageMaker endpoint.