Machine Learning Engineer and AI Solutions Architect with 20 years of technology experience, including 8 years across classical machine learning, deep learning, computer vision, generative AI and Agentic AI, backed by 12…
Machine Learning Engineer and AI Solutions Architect with 20 years of technology experience, including 8 years across classical machine learning, deep learning, computer vision, generative AI and Agentic AI, backed by 12 years of performance-critical C++ engineering. Designs and implements ML and LLM solutions spanning data preparation, model training and validation, LoRA/QLoRA fine-tuning, retrieval, model serving and cloud deployment using Python, PyTorch, TensorFlow, Hugging Face, FastAPI, Docker, Azure and AWS. Combines hands-on engineering with ML system architecture, secure API integration, performance optimization and production-readiness design.
<p>Developed a Python connector in Azure Databricks to extract Collibra metadata through REST APIs, transform and map it to Glean's indexing schema, and ingest it through the Glean Indexing API.</p><p> Implemented secure, read-only Collibra access using functional account credentials with Basic Authentication, with secrets managed through Azure Key Vault.</p><p> Enabled governed Collibra content discovery through Glean enterprise search and downstream RAG for the client's enterprise chatbot.</p><p> Currently working on a permission-aware InfoDocs-to-Glean integration.</p><p> Built a multi-agent workflow that interprets SAP manual test cases, produces validated structured JSON and transforms it into executable Tricentis Tosca XML scripts.</p>
<p>Developed AVEVA Process Optimization components for real-time and offline oil and gas process modelling using C++, VC++, MFC and COM.</p>
<p>Developed Fiery JDF communication components connecting print-workflow applications with Fiery Print Servers using C++, MFC and COM.</p>
<p>Enhanced LaserJet current-product engineering and firmware-download utilities using C++, VC++ and Win32 APIs.</p>
<p>Developed AXIS medical-imaging components and DICOM clinical workflows on the Siemens Syngo framework using C++, VC++ and MFC.</p>
<p>Developed INSITE engine diagnostics for parameter configuration, fault-data analysis and ECM troubleshooting using C++ and VC++.</p>
<p>Built a document-intelligence pipeline combining Regex and Mistral inference to extract invoice numbers, dates, amounts and other financial entities from unstructured email and tabular content.</p><p> Containerized the service with Docker and deployed Mistral inference on AWS EC2 GPU for model serving; integrated Amazon S3 for document exchange and designed an AWS Lambda trigger for event-driven processing.</p>
<p>Built a domain-specific plant-engineering assistant using a hybrid PEFT + RAG design for manuals, SOPs, maintenance logs and troubleshooting guides.</p><p> Fine-tuned LLaMA-family models with LoRA/QLoRA, grounded responses with Sentence Transformers and Pinecone, exposed the service through FastAPI, and packaged it with Docker for Kubernetes deployment.</p><p> Built CNN and EfficientNet rust/non-rust image classification for chemical-plant equipment; generated synthetic rust images with Stable Diffusion to augment scarce training data and exported the model through ONNX.</p><p> Built and evaluated a customer-matching workflow across 2.5 million financial records and 39,009 customer classes, comparing KNN with FAISS approximate nearest-neighbor search using Recall@5.</p><p> Developed an interpretable supervised ML workflow on structured plant-operational data to distinguish equipment faults from control-system issues, using Logistic Regression, feature engineering, cross-validation, ROC-AUC and confusion-matrix analysis.</p>
نظام التوظيف التشغيلي المعتمد على الذكاء الاصطناعي أولاً، الذي يوحّد ويؤتمت دورة التوظيف بالكامل للمؤسسات الحديثة حول العالم.
د.جوب منطقة حرة ذ.م.م. 2026 © جميع الحقوق محفوظة
