Skip to content
Marwan AbdElhameed's profile pictureMarwan AbdElhameed

Projects

Research artifacts, open-source tooling, and production systems.

research

AI-Powered Requirements Analysis Platform

  • Architected intelligent SRS analysis platform integrating 3 LLM APIs with advanced aggregation strategies, achieving 95% consensus accuracy.
  • Engineered document processing pipeline supporting PDF/DOCX/TXT with Mistral AI OCR integration.
  • Developed UML domain model generator reducing manual modeling time by 80%.
  • Implemented requirements completeness analyzer improving SRS quality by 60%.
Python
Flask
JavaScript
Multi-LLM APIs
research

Math Problem-Solving LLM System

  • Fine-Tuned Mistral-7b achieving 35% accuracy on GSM8K dataset.
  • Enhanced solution accuracy by 25% from base model through advanced inference scaling techniques.
  • Created systematic evaluation framework for optimal prompting and inference strategy selection.
Python
Large Language Models
research

Distributed Model Serving Platform

  • Built scalable REST API handling concurrent model inference requests, achieving 99% uptime.
  • Integrated Ray with TorchX for distributed computing, processing 10,000+ concurrent requests.
  • Implemented efficient task distribution and monitoring system, reducing inference latency by 20%.
Flask
Ray.serve
TorchX
research

Multilingual RAG System

  • Architected multilingual RAG system with mBART-large-50 for cross-lingual content generation.
  • Improved content relevance by 40% through context-aware document fetching.
  • Scaled system to handle 100k+ document pairs.
Python
mBART
Vector DB
research

Distributed ML Training System

  • Developed distributed training system using PyTorch's DataParallel, achieving 1.92x speedup with 4 GPUs.
  • Optimized batch size scaling to achieve 31,063 samples/second throughput (a 7.3x improvement).
  • Reduced I/O bottlenecks by 40% through custom data loading pipeline.
PyTorch
CUDA
Parallel Computing
research

ML Infrastructure Optimization

  • Architected batch prediction system using Ray actors and TorchX, processing 250 requests per batch.
  • Implemented multi-worker data loading and distributed hyperparameter optimization.
  • Developed threaded execution pipeline using Python's threading and ZMQ, reducing model serving latency by 28%.
Python
Ray
TorchX
W&B
research

Seq2Seq Conversational AI

  • Engineered sequence-to-sequence model processing 220,000+ movie dialogue exchanges with nn.GRU and nn.Embedding.
  • Implemented Luong attention mechanism, improving response relevance by 30%.
  • Developed custom text preprocessing pipeline for 18,000+ unique vocabulary tokens.
PyTorch
NLTK
Neural Networks
research

SSD ONNX Model Integration

  • Engineered computer vision model achieving 93.33% accuracy in object detection using MobileNet V1 SSD.
  • Optimized model performance through hyperparameter tuning to 94.31% validation accuracy.
  • Reduced inference time by 25% through ONNX model conversion.
PyTorch
ONNX
Computer Vision
personal

NYT Articles Analysis Pipeline

  • Engineered scalable data pipeline processing 100,000+ NYT articles monthly with 98% accuracy.
  • Developed text analytics system using NLTK and Gensim, reducing processing time by 75%.
  • Architected end-to-end ML pipeline with LDA for topic modeling.
Python
Data Analysis
NLP
personal

Photo Sharing Web Application

  • Built full-stack app with user authentication, photo upload, and album management, increasing retention by 25%.
  • Optimized data relationships and queries, improving retrieval efficiency by 40%.
  • Implemented social features (friends, tagging, commenting), boosting user engagement by 30%.
Python
Django
SQL
PostgreSQL
personal

Personal Finance Manager

  • Developed finance management web app with interactive charts, increasing user engagement by 40%.
  • Implemented efficient data management with PyMongo, reducing query response time by 50%.
  • Deployed using Docker and DigitalOcean with CI/CD pipeline, ensuring 99.9% uptime.
Python
Flask
PyMongo
Docker
personal

What to Eat Today?

  • Launched meal suggestion app, attracting 1,000+ active users in three months.
  • Integrated with Spoonacular API, offering 5,000+ recipes with 99.9% uptime.
  • Implemented AJAX for real-time updates.
JavaScript
Node.js
Express.js
MongoDB

Other repositories

Recently pushed public repositories, straight from GitHub.

WhatToEat

No description available.

0
0
JavaScript

thriftthelook

Turn one outfit photo into a secondhand look you can actually buy under one all-in budget.

0
0
Python

Debris Hunters

Project's page in Nasa Space Apps official website

game
game-development
+2
0
0
C#

Brawndo

python-package-exercise

0
1
Python

ASE 2025_Software Requirements Completeness Analyzer

A software requirements analysis tool that uses multiple LLMs (DeepSeek, OpenAI, Claude) to generate domain models, detect missing requirements, and analyze requirement completeness with intelligent aggregation strategies.

0
1
Python

xarmvoicecontrol

A code that controls Xarm6 with integration with Gemini voice API for chat and command understanding

0
1
Python
See more on GitHub →