AI Engineering Student

Joined the Quality Perfection team in Riyadh as a QA Engineering Intern to strengthen validation and test automation on the Tamam SaaS platform.

• Documented over 80 defects through API automation, AI-generated test scenarios, and UI automation on the Tamam SaaS platform, each reproducible with evidence and criticality.
• Highlighted systematic failures in LLM-generated test cases: hallucinated assertion data, fabricated field names, incorrect boundary values, off-by-one errors, and non-reproducible scenarios.
• Traced root causes beneath the interface through direct API calls and network-level inspection, isolating blocking defects in request serialization, authentication, and chaining.
• Developed a Python CLI validating OpenAPI, Swagger, and Postman collections while masking credentials, covered by 44 automated tests.
• Technical environment: Python, Postman, Swagger/OpenAPI, WSDL/SOAP, Robot Framework, Jira, REST APIs

Joined the IT, Data Factory division at Royal Air Maroc in Casablanca to deliver three decision support tools in eight weeks.

• Modeled a 13-table MySQL database with 21 analytical queries covering punctuality, fleet, commercial activity, crew conflicts, and data quality.
• Built a Power BI dashboard (DAX measures, inactive relationships, generated date table) surfacing operational KPIs for management.
• Automated a flight weather enrichment pipeline in n8n, crossing three external sources via parallel branches into a structured CSV export.
• Developed a document summarization agent on a self-hosted Llama 3.2 model (Flask, Ollama), running entirely offline to meet a strict confidentiality constraint, and validated each automated output against an independent source.
• Technical environment: MySQL, SQL, Power BI, n8n, Python, Flask, Ollama, REST APIs, Git

Joined Shared System in Casablanca to develop data cleaning solutions and artificial intelligence pipelines.

• Developed a Python AI agent to clean and normalize Port State Control maritime inspection datasets, resolving missing values, duplicates, format inconsistencies, and entry errors.
• Integrated machine learning models for anomaly detection across inspection records.
• Industrialized processing behind Flask and FastAPI REST APIs, with a Streamlit interface presenting results in real time.
• Built a fully self-hosted voice pipeline for Arabic, English, and the Saudi dialect: voice activity detection, speaker diarization (NeMo Sortformer, ECAPA-TDNN), language identification, and language-routed ASR across Cohere and Qwen.
• Measured quality with WER and DER, and ran oracle-diarization experiments to separate diarization error from ASR error on overlapping speech, re-injecting results into glossary correction and decoding parameter tuning.
• Transformed the pipeline into live voice agents for a drive-thru and a hospital line, with Qwen served on vLLM, an MCP server exposing tools on a FastAPI REST API, and streaming text-to-speech.
• Kept business logic out of the model: pricing and bookings are calculated in code, with booking integrity ensured by database constraints and locking, proven by a 20-thread concurrency test.
• Reduced conversational latency from a peak of 16 seconds to 2.4–4.6 seconds per turn by instrumenting each step and fixing a query cache threshold effect; 46 out of 46 hospital scenarios passed, supported by 36 unit and concurrency tests.
• Technical environment: Python, Scikit-learn, Pandas, Flask, FastAPI, Streamlit, PyTorch, NeMo, Cohere, Qwen, vLLM, SQLite, REST APIs

Completed an internship at VNB IT in Paris focused on web interface development and data integration.

• Translated design specifications into pixel-perfect React.js interfaces for a production driving school application, connecting them to backend REST APIs.
• Developed lesson planning and calendar navigation for students and instructors, and rebuilt multi-format data export (Excel, CSV) for admin, teacher, and student roles.
• Functionally tested every feature across the three user roles prior to delivery.
• Technical environment: React.js, REST APIs, Excel, CSV