Ahmed Taha

AI & Automation Engineer
Ismailia, EG.

About

Highly innovative AI & Automation Engineer with a proven track record of building and deploying production-grade AI tools, agents, and automated workflows. Expert in LLM systems, computer vision, and advanced cloud stacks (GCP/AWS/Azure), consistently delivering solutions that enhance efficiency, accuracy, and scalability. Driven to leverage deep technical expertise to solve complex challenges and drive significant technological advancements in dynamic environments.

Work

Self-Employed
|

Freelance AI Engineer & Automation Specialist

Remote

Summary

Provided specialized AI engineering and automation consulting, developing advanced LLM-based agents and production-level bots for diverse client needs.

Highlights

Conducted advanced research on LLM safety using malicious Arabic prompts, rigorously testing SOTA models to enhance safety alignment and robustness.

Developed a sophisticated retail support agent utilizing LangGraph and DSPy, enabling efficient database querying via SQL for improved customer service operations.

Built and deployed a production-level file distribution bot, reliably syncing local folders to specific clients on WhatsApp, enhancing communication and data dissemination efficiency.

Agroptics
|

Founding AI Engineer

Remote

Summary

Led the end-to-end development and DevOps lifecycle for a greenfield geospatial analytics system, ensuring robust data processing and optimized cloud infrastructure.

Highlights

Architected and built a greenfield geospatial analytics system from conception to deployment as a solo engineer, delivering full-stack functionality.

Engineered solutions for out-of-order, concurrent, and overlapping data ingestion, significantly improving data integrity and processing efficiency.

Implemented a stateless ingest and runtime aggregation architecture with a cache layer, enhancing system performance and scalability.

Managed the complete DevOps lifecycle, optimizing cloud compute infrastructure and automating data pipelines, resulting in improved resource utilization and faster deployments.

Maintained and improved a legacy production system, proactively resolving issues and patching flaws to ensure continuous operation and reliability.

Education

Faculty of Engineering, Suez Canal University
Ismailia, Ismailia, Egypt

Bachelor

Computer Engineering

Grade: GPA: Very Good | Graduation Project Grade: Excellent

Publications

A Comprehensive Approach to Arabic Handwriting Recognition: Deep Convolutional Networks and Bidirectional Recurrent Models for Arabic Scripts

Published by

The International Journal of Telecommunications (IJT)

Summary

Authored a publication detailing a comprehensive approach to Arabic handwriting recognition using deep convolutional networks and bidirectional recurrent models.

Languages

English
Arabic

Skills

Languages & Frameworks

Python, TypeScript, FastAPI, React, Next.js.

AI/ML

LLMS, RAG, LangChain, PyTorch, TensorFlow, Computer Vision, NLP, Gemma, Gemini.

Data & Infra

SQL, PostgreSQL, Redis, Docker, Cloud (GCP/AWS/Azure), Qdrant, SQLite, Cloudflare Workers.

Projects

Searchable Screenshots, Local AI-Powered Image Search

Summary

Engineered a local AI-powered image search system, combining hybrid retrieval with advanced OCR and AI-driven description generation.

DeWallify, AI-Powered Markdown Formatter

Summary

Built a web application that transforms unstructured text into clean, semantic docx format, leveraging orchestrated AI models.

Focal, AI-Powered Financial Tracker

Summary

Developed a full-stack AI-powered financial tracker with secure authentication and advanced expense extraction capabilities.

Knitty, AI Resume Optimizer

Summary

Created a multi-agent LangChain system to optimize CVs against job postings, integrating various data sources and achieving significant matching improvements.

Digital-Dad الرقمية - Arabic Handwriting OCR System

Summary

Developed a robust Arabic handwriting OCR system and an AI agent for automated exam grading, significantly improving processing efficiency.