Wa'el Mohsen

Architectural Modernization and Integration

• Captured and analyzed current architectures to integrate, harmonize, and modernize legacy applications into unified modern architectures.
• Created comprehensive architectural blueprints outlining integration strategies, key components, interfaces, and data flow.
• Identified integration opportunities, defined patterns, and developed strategies to modernize legacy systems, ensuring data consistency, interoperability, and seamless communication.
• Used Cursor and GitHub Copilot for AI-assisted code reviews, enforcing quality gates and refactoring consistency across vendor teams.

AI and Data Strategy

• Defined the vision and adoption strategy for data and AI aligned with company goals, leveraging technologies like Azure AI services, GenAI technologies, and OpenAI foundation models.
• Designing & Leading the development of LLM-based claims medical translation platform, travel insurance customer support agent, and sales support platform, incorporating RAG architectures, Prompt Engineering, Azure AI Foundry, AutoGen and Semantic Kernel, Azure cognitive services, Azure integration services, Azure storage & database services, and OpenAI foundation models.
• Designing and implementing evaluation techniques for LLM-based applications, focusing on precision, recall, and accuracy metrics while leveraging tools like Hugging Face, LangChain, and Azure AI services to assess model performance, scalability, and cost-effectiveness.
• Designed and deployed multi-agent LLM orchestration systems using LangChain, AutoGen, and Semantic Kernel for document processing and financial data pipelines.

• Medical Claim Translation & Eligibility Agent – Designed and implemented an agentic AI pipeline using MS Semantic Kernel to automate medical claim processing. The system extracts text from medical claims, translates it into English, and applies domain-specific business rules to validate eligibility, ensuring compliance with HIPAA standards and significantly reducing manual review time.

• Dynamic Fraud Detection Workflow – Developed a rule-driven agentic workflow powered by Semantic Kernel to detect fraud in claims. The system orchestrates multiple agents and tools to apply configurable business rules dynamically, flagging high-risk claims for further investigation and improving fraud detection accuracy.

• Utilized advanced developer tools, including VS Code’s Cline extension, Cursor, and Trae, to enhance productivity and streamline development workflows—leveraging Cline for AI-assisted code generation, Cursor for intelligent code navigation and refactoring, and Trae for efficient API testing and integration within GenAI-powered applications.
• Ensure compliance with GDPR, HIPAA, and CCPA standards.

Legacy to Microservices Modernization

• Modernized legacy monolithic applications into scalable, cloud-ready microservices by analyzing codebases, identifying modularization opportunities, applying Domain-Driven Design (DDD) principles, and implementing robust microservice architectures with CI/CD pipelines, Azure DevOps, and effective communication patterns (REST APIs, asynchronous messaging).
• Implemented Azure DevOps CI/CD pipelines for AI model deployment, integrating unit tests, containerization, and Azure Functions triggers.
• Modernize applications from (.Net Framework 3.5) to (.Net 9), revamping applications from ColdFusion, Java to .Net 9 and Blazor-based Apps, and introducing Python to build GenAI Solutions.

Software Architecture and Tool Utilization

• Evaluated and integrated tools like NDepend, SonarQube, and Fortify for static and code analysis, ensuring secure, maintainable, and high-quality codebases.
• Designed and diagrammed software architectures using tools like Visio and Miro, simplifying complex technologies for business leaders.
• Enforced coding standards and security policies using tools like Fortify and conducted load and stress testing for scalable applications.
• Applied expertise in SQL and NoSQL to enhance system architecture and optimize application performance.

API Development and Optimization

• Architected, designed, and developed REST & gRPC APIs, OpenTelemetry, Seq, RabbitMQ ensuring high performance and scalability.
• Build an integration layer to integrate with multiple service providers, such as Zesty CMS, Email services, SMS services, Okta SSO providers, and various GenAI providers (OpenAI, Anthorpic, Hugigng Face).

• News and customer feedback clustering and topic modeling projects using BERTopic, embeddings, and vector databases.
• Enterprise Recommendation Systems for Ministry Services Using Classical and then GenAI Techniques. Techniques included Embedding Models, dimensionality reduction, autoregressive and representation models, topic modeling, and MLOps.
• Arabic Sales & Orders Chatbot for retail food products using RAG, LlamaIndex, CromaDB, LangChain, Pipelines, NER, Synthetic data, data exposure, data normalization, AraBERT, OpenAI APIs, Cohere Re-ranking model, Term-based and Embedding-based similarity search, and LangSmith.
  • Claims Medical translation Platform using Azure AI Foundry, Azure Document Intelligence, Azure OpenAI, Azure Functions, Event Grid, Cosmos DB, Azure AI Search, Azure Monitoring & Insights.
  • Ensure compliance with regulatory requirements and internal risk management standards. and the NIST Responsible AI framework along with HIPAA & GDPR.
• Travel Insurance Customer Support Agent using RAG, LangGraph, LangSmith, and VS Code and Cursor.
• Agentic AI App to write/post Multi-lingual Posts/articles and post it to LinkedIn. This agentic system crawling the internet for the latest news in a specific set of fields, judge the news importance and affect, get enough details about the filtered news, then write English post, translate it to different language(s), and finally either save the posts/articles to the HardDisk or post it to LinkedIn. This app uses Semantic Kernal, MCP, crawling and internet search tools, Cline (AI-powered coding assistant that runs in VS Code) plugin to Visual Studio Code, Lang Fuse, and custom tools to post to LinkedIn.
• Sales support Agents Crew using MS Semantic Kernel Agentic AI framework, OpenAI APIs, MCP, Tools, LangFuse, RAG, VS Code and Trae.
• Implemented advanced generative AI pipelines using PyTorch, Hugging Face Transformers, and LangChain, ensuring robust data governance and security at each stage of model development.
• Fine-tuning initiatives for large language model on domain-specific tasks (Biology field and General field) and optimizing accuracy while enforcing strict data security and compliance standards. We used PEFT/LoRA & Knowledge Distillation, and RLHF.
  • Applied Explainable AI (XAI) with LIME and SHAP for model interpretability. Ensuring the transparency and fairness in AI-driven decisions.

• Expert Systems for Plant Medical Diagnosis.
• Multiple OCR & ICR projects incorporating Image processing & AI techniques.
• CNN & RNN for multiple biometric projects such as face recognition, Iris recognition, and fingerprints.
• ML decision support system using supervised regression techniques and unsupervised clustering techniques.
• Enterprises Recommendation systems for ministry services.