Top Skills
About Me
I'm an AI/ML Engineer with about 3 years of experience building production systems across e-commerce, conversational AI, and NLP.
My work focuses on turning raw, unstructured data into intelligent solutions — from improving product discovery and search quality to developing scalable AI pipelines and multimodal systems. I've worked on projects that improved image-text matching accuracy by 50% and increased data pipeline efficiency by 300%, directly impacting user experience and system performance.
I have hands-on experience with multimodal AI (image-text matching, vision-language models), RAG systems and AI agents, fine-tuning language models in Arabic and English, text-to-speech systems, and end-to-end AI pipelines from data to deployment.
Experience
Applied Scientist
Microsoft · Cairo, Egypt
- Designed AI-driven pipelines to enrich large-scale e-commerce data with high-quality product images
- Fine-tuned BLIP using contrastive learning, achieving 50% improvement in image-text matching accuracy
- Partnered with the Copilot team to integrate Shopping functionality into the Copilot platform
Data Scientist
WideBot · Cairo, Egypt
- Automated text-to-speech data creation and cleaning pipelines, boosting production capacity by 300%
- Fine-tuned TTS models (XTTS V2, fish-speech) for Egyptian dialect generation in AI agents
- Built a fast Arabic text summarizer in MSA and Egyptian dialect, achieving 94% similarity score
Coding Instructor
iSchool · Remote
- Taught Python and basic AI concepts to over 80 students under the age of 13
AI Developer — Summer of Reproducibility
Flower Labs · Remote
- Built an open-source federated learning system replicating results from a published paper on Gradient-less Federated Gradient Boosting Trees
- Used Flower framework, Hydra-config, W&B, Poetry, PyTorch, and XGBoost
Junior ML Engineer
Omdena · Remote
- Collaborated with cross-functional teams to build a chatbot for providing free therapy
- Cleaned text data using NLP algorithms and clustered text data based on semantics
"She implemented a state-of-the-art federated AI algorithm demonstrating not only her coding skills but also her ability to understand the recent research literature. She is extremely hard working, a fast learner, and has demonstrated that she can deliver projects to an outstanding level of quality."
Publications
Gumball at QIAS 2025 Shared Task: Arabic LLM Automated Reasoning in Islamic Inheritance
ArabicNLP Conference 2025, China
Publisher: Association for Computational Linguistics (ACL)
Read Paper →Education
Bachelor of Computer Engineering
Faculty of Engineering (Shoubra), Benha University
2018 — 2023 · Grade: Very Good (80.53%)
Graduation Project: Fraud Detection System Using Federated Learning — 98% accuracy, 93% AUC · Grade: Excellent
GET IN TOUCH
- aml.hassan.esmil@gmail.com