I'm an AI Engineer and independent AI researcher focused on building systems from model fundamentals to production infrastructure.

My work spans:

• Generative AI & LLM engineering — RAG, agents, tool use, inference, fine-tuning, evaluation, and model architecture.
• AI systems engineering — APIs, orchestration, multi-tenant SaaS, streaming, observability, and production workflows.
• Model research — studying and implementing modern architectures such as Titans, MIRAS, MoE systems, long-context methods, and efficient attention/memory mechanisms.
• Applied AI — healthcare, pharmaceutical quality systems, NLP, computer vision, and multimodal applications.
• AI infrastructure — Docker, CI/CD, GitHub Actions, cloud deployment, model serving, and reproducible engineering.

I'm currently completing a B.Sc. in Electronics & Communications Engineering at Helwan University (expected 2027).

I care about understanding the system underneath the abstraction—not only calling an API, but understanding the model, data pipeline, runtime, and infrastructure around it.

An open-source Arabic/English LLM research project focused on building the stack rather than treating the model as a black box.

Current direction:

• Custom tokenizer and bilingual data pipeline
• Arabic + English pretraining
• Transformer and memory/attention research
• Titans / MIRAS-inspired experiments
• MoE experimentation
• Efficient training and inference
• Long-context research
• Evaluation and reproducible experiments

Research direction: build progressively from a text foundation model toward multimodal capabilities.

→ ATHLLM

A unified platform for building AI coding agents and AI-native developer products.

The architecture is intended to support:

• CLI
• IDE integrations
• Web applications
• Coding agents
• Design agents
• Cowork-style workflows
• MCP/tool integration
• Multi-provider routing
• BYOK provider connections
• Streaming
• Sessions and context
• Guardrails
• GitHub workflows and CI/CD
• APIs for building AI applications on top of the platform

The goal is not to build another thin chat wrapper. The goal is to provide the runtime and engineering foundation around coding agents.

→ Ancient

A multi-tenant quality and stability management platform for pharmaceutical organizations and laboratories.

Engineering focus includes:

• Organization/site isolation
• RBAC and object authorization
• Stability studies and protocols
• Controlled results and approval workflows
• Audit trails and immutable records
• OOS / OOT / Deviation / CAPA / Change Control
• Reporting and exports
• Subscription and entitlement architecture
• Paymob payment integration
• PostgreSQL, Redis, Celery, Docker and Nginx
• Production CI/CD and release gates

QCSTS is positioned as designed for GxP-regulated environments with a validation-ready architecture, not as automatically certified regulatory software.

→ QCSTS

I also build healthcare-oriented AI systems covering:

• Medical imaging
• Clinical decision-support prototypes
• Multimodal AI
• Medical NLP
• Healthcare data pipelines
• AI-assisted hospital systems

The emphasis is on combining AI with real software architecture, rather than building isolated notebooks.

→ Medical AI Platform

• Titans
• MIRAS
• Mixture-of-Experts architectures
• Long-context architectures
• Modern open-weight LLMs
• Efficient attention and memory mechanisms
• Retrieval-augmented generation
• Agentic software engineering

Languages

AI / ML

PyTorch · Transformers · Hugging Face · Scikit-learn · RAG · LangChain · LangGraph · FAISS · Vector Databases

Backend & Systems

Engineering & Infrastructure

CI/CD · Docker Compose · Celery · MLflow · REST APIs · MCP · Observability · Production Testing

• Building RAG systems with retrieval, reranking, context construction, and evaluation.
• Designing coding-agent runtimes rather than only prompt-based assistants.
• Working with multiple model providers and local inference.
• Fine-tuning and experimenting with open-weight models.
• Studying model internals and implementing architectural ideas from research papers.

• Designing multi-tenant SaaS architectures.
• Building API-first AI systems.
• Implementing authorization and tenant isolation at the backend boundary.
• Containerizing services and supporting production deployment.
• Building CI/CD pipelines with automated tests and release gates.
• Integrating AI into domain-specific business workflows.

• NLP classification and sequence models.
• Computer vision and object detection.
• Medical imaging.
• Feature engineering and model evaluation.
• Model compression and inference optimization.

Understand the abstraction
        ↓
Measure the system
        ↓
Design the boundary
        ↓
Implement the smallest correct primitive
        ↓
Test failure modes
        ↓
Automate the workflow
        ↓
Deploy reproducibly
        ↓
Observe and iterate

I prefer systems that are:

• Explicit over magical
• Testable over optimistic
• Observable over opaque
• Reproducible over environment-dependent
• Secure by architecture over convention alone
• Simple to understand without sacrificing capability

B.Sc. Electronics & Communications Engineering
Helwan University · Expected 2027 · Egypt

Relevant areas:

Machine Learning · Deep Learning · NLP · Computer Vision · Signal Processing · Pattern Recognition · Embedded Systems

AI Engineering
     │
     ├── Production LLM Systems
     │      ├── RAG
     │      ├── Agents
     │      ├── Tool Use
     │      └── Evaluation
     │
     ├── Model Research
     │      ├── ATHLLM
     │      ├── Memory / Attention
     │      ├── MoE
     │      └── Long Context
     │
     ├── AI Infrastructure
     │      ├── Containers
     │      ├── CI/CD
     │      ├── Model Serving
     │      └── Cloud
     │
     └── AI Products
            ├── Developer Tools
            ├── Healthcare
            └── Pharmaceutical Systems