-
🌱 I’m currently pursuing: Machine Learning Engineering, Research Engineering.
-
🛠 I've experienced Freelancing, Online Tutoring, and working on Machine/Deep Learning, NLP, and LLMs projects as a contractor.
-
🧑🏻💻Currently experimenting with PEFT algorithms in domains other than NLP like Computer Vision applications, GenAI applications (recommendation Sys, RAGs, ..etc.), benchmarking and evaluation systems for domain-specific problems, and representation systems for different types of data."To the vector store, or the knowledge graph, maybe both, maybe another novel representation".
-
🧠 Interests:
-
Domain-Specialized AI Development
Embedding deep problem-domain knowledge into model architectures and workflows to maximize efficiency and minimize energy footprint. -
PEFT Beyond NLP
Applying LoRA/QLoRA/DoRA-style adapters in Computer Vision, GenAI (recommendation systems, RAG pipelines), and other modalities. -
Benchmarking & Evaluation Systems
Designing end-to-end suites for rigorously comparing domain-specific models (accuracy, latency, resource cost). -
Novel Representation Learning
Exploring vector stores, knowledge graphs, and hybrid/graph-neural approaches for rich, flexible knowledge encoding. -
Interpretability & Alignment
Probing model decision paths, emergent behaviors, and safety mechanisms (long-term Anthropic-style inquiry). -
AI ↔ Cybersecurity Integration
Securing ML pipelines (privacy, robustness) and embedding AI into security tooling (threat detection, anomaly analytics). -
3D Avatar & VRoid-Based Services
Automating avatar generation workflows, integrating real-time facial motion capture (VSeeFace/VMagicMirror) with AI -
Computational Neuroscience & Brain-Signal ML
Real-time joint/site detection; EEG/fMRI-driven models for harmful-activity prediction in medical contexts.
-
📝 Blog | 🐦 X | 🤝 LinkedIn
- 📫 How to reach me :
- 📧 Personal / Business contact : moh.z.ahmed007@gmail.com
- 📧 Academic contact: s-mohamedzayed@zewailcity.edu.eg