ONCF · Internal Audit · Rabat
Software Engineering Intern
Developed and integrated two modules of AuditFlow, an internal audit management platform.
CASE STUDYAuditFlowSOFTWARE ENGINEER / APPLIED AI
Chemlali Ibtihal
I build thoughtfulsoftware that connectspeople, systems& intelligence.
Final-year Software Engineering & Intelligent Systems student at ENSAM Meknès. A curious mind, a considered approach, and a desire to build things that matter.
Open to a 5–6 month PFE internship · 2027
ENSAM MEKNÈSCuriosity → Design → CodeFRONTENDBACKENDARTIFICIAL INTELLIGENCESOFTWARE ARCHITECTURE
01Behind the work
A curious mind.
An engineering lens.
I’m Ibtihal Chemlali, a software engineering and intelligent systems student. I enjoy turning complex needs into practical applications—from enterprise audit workflows to accessible communication through computer vision.
My interests sit at the intersection of reliable software, thoughtful user experiences and applied AI. I’m seeking a 5–6 month final-year internship for 2027 to contribute, learn and work on meaningful engineering challenges.
01Thoughtful interfaces
02Connected systems
03Intelligent applications
Education
2022 — 2027
ENSAM Meknès
Engineering cycle · Software Engineering & Intelligent Systems
5th-year engineering student2020 — 2021
Mathematical Sciences Baccalaureate, option A
Lycée Imzouren
LEARN / BUILD / CONNECT
02Field notes
Learning from
the real world.
Professional experience translating business needs into working software.
2026
01 JUL — 20 AUG 2026
2025
JUL — AUG 2025
Génération Maroc Technologie · Rabat
Software Development Intern
Built a commercial management application for sales, purchases, inventory, customers and suppliers. Designed REST APIs, role-based JWT authentication, persistence and PDF/Excel document generation.
CASE STUDYGestion Commerciale03Selected work / 2025 — 2027
Engineering,
in practice.
Five projects. Different contexts. The same attention to the people, the systems and the details behind them.
Professional / Enterprise software
/ 05AuditFlow
ONCF · 01 JUL — 20 AUG 2026
Bringing structure to complex audit workflows.
MY CONTRIBUTION
During my ONCF internship, I developed and integrated two major modules in an existing internal audit platform.
- Audit entities, risk evaluation and annual planning.
- Resources, assignments and approval workflows.
- Mission preparation, QCI, TREC and audit testing programs.
Experimental AI assistance. Professional source code and internal documents remain confidential.
BUILT WITH
Architecture flow
Technical architecture
Hover, focus or tap a component to explore its role.
AuditFlow is the platform I worked on during my ONCF internship. Two distinct server-side paths reach the experimental AI services; the browser never calls FastAPI directly.
Application / data pathExperimental AI path
Verified component connections
- Web interface → Next.js API routesHTTP · server route
- Web interface → Business APIREST API
- Next.js API routes → Python / FastAPI servicesServer-side HTTP
- Business API → Python / FastAPI servicesServer-side HTTP
- Business API → Business persistenceEF Core · SQL
- Python / FastAPI services → AI toolsGeneration / research
- Python / FastAPI services → Intermediate cacheSelected AI processing
Technology stack
Frontend
- Next.js
Web framework and server-side API facade routes.
- React
Audit workflow interfaces and reusable components.
- TypeScript
Typed frontend application code.
- Material UI
User-interface component system.
Backend
- ASP.NET Core Web API
Business APIs and server-side AI orchestration.
- Entity Framework Core
Relational persistence through the backend.
Databases & persistence
- PostgreSQL
Relational business data.
- JSON cache
Intermediate cache for selected AI treatments.
AI & experiments
- Python
Implementation of experimental AI services.
- FastAPI
HTTP service endpoints reached only from server-side callers.
- Groq
Generation used by the AI service layer.
- Tavily
Documentary research used by AI services.
Applied AI / Accessibility
/ 05ASL Translator
Academic project · 2026
AI-powered American Sign Language recognition and communication.
MY CONTRIBUTION
I worked on static and dynamic sign recognition models and their integration into a communication-focused web application.
- ResNet50 transfer learning for 39 static sign classes.
- MediaPipe landmark sequences with BiLSTM recognition.
- Recognition integrated with posts, messaging, stories and video calls.
92.57%ResNet50 · experimental accuracy94.72%BiLSTM · experimental accuracy
Reported experimental model results. Evaluation protocols were not provided; these figures do not establish production performance.
BUILT WITH
Architecture flow
Technical architecture
Hover, focus or tap a component to explore its role.
Two separate model experiments feed application recognition. ResNet50 classifies static images; MediaPipe landmark sequences feed BiLSTM for dynamic signs. The models are not a single sequential pipeline.
Application / data path
Verified component connections
- Static sign images → ResNet50 experimentModel inference
- ResNet50 experiment → Static predictionClassification
- Static prediction → Application integrationIntegration
- Gesture sequences → MediaPipe landmarksLandmarks
- MediaPipe landmarks → BiLSTM experimentLandmark sequences
- BiLSTM experiment → Dynamic predictionSequence inference
- Dynamic prediction → Application integrationIntegration
Technology stack
Frontend
- React
Web integration of recognition and communication.
AI & experiments
- Python
Model experiments and landmark processing.
- TensorFlow / Keras
Deep learning model implementation.
- MediaPipe
Landmark extraction for the dynamic sign branch.
- ResNet50
Separate static sign experiment; accuracy 92.57%.
- BiLSTM
Separate dynamic sign experiment; accuracy 94.72%.
Professional / Business systems
/ 05Gestion Commerciale
GMT Maroc internship project · Jul–Aug 2025
A commercial management application developed during my GMT Maroc internship.
MY CONTRIBUTION
I developed the commercial management application during my GMT Maroc internship, including business APIs, persistence and role-based authentication.
- Sales, purchases, inventory, customers and suppliers.
- REST APIs and relational persistence with Spring Data JPA/Hibernate.
- JWT role-based authentication and PDF/Excel document generation.
BUILT WITH
Architecture flow
Technical architecture
Hover, focus or tap a component to explore its role.
A React client connects to a layered Spring Boot API for sales, purchasing and inventory. MySQL stores business data; the Nova assistant calls Groq through the backend.
Business capabilitiesSalesPurchasesInventoryCustomers & suppliersPDF / Excel documents
Application / data path
Verified component connections
- Commercial interfaces → Business APIREST · JSON · JWT
- Business API → Relational persistenceJPA / Hibernate · SQL
- Business API → Nova assistantChat requests
- Nova assistant → Groq APIServer-side HTTP
Technology stack
Frontend
- React
Interfaces for sales, purchases, stocks, customers and suppliers.
Backend
- Spring Boot
Commercial business APIs and document generation.
- Spring Data JPA / Hibernate
Persistence and object-relational mapping.
Databases & persistence
- MySQL
Relational storage for the application.
Security & AI
- JWT
Access and refresh tokens with Spring Security.
- Groq
Nova assistant, called from backend services.
Full-stack / Education
/ 05QuizArt – Online Examination Platform
Academic project · 2025–2026
Connecting ENSAM teachers and students through online exams.
MY CONTRIBUTION
I developed an examination platform connecting ENSAM teachers and students through secure online exam workflows.
- Teachers create and manage exams and quizzes.
- Students take exams online and access their examination dashboard.
- Secure authentication, hybrid data architecture and asynchronous submissions.
BUILT WITH
Architecture flow
Technical architecture
Hover, focus or tap a component to explore its role.
A React application for students, professors and administrators, backed by a layered Spring Boot application. MySQL and MongoDB store complementary data, while RabbitMQ processes submissions asynchronously.
Business capabilitiesExam authoringTimed assessmentsProgramming questionsGrading & progress
Application / data path
Verified component connections
- Web application → Secured REST APIHTTP · JSON · JWT
- Secured REST API → Application servicesBusiness calls
- Application services → Relational dataJPA · SQL
- Application services → Document dataDocuments
- Application services → Account emailSMTP
- Application services → Submission queuePublish
- Submission queue → Submission consumerAMQP delivery
- Submission consumer → Application servicesProcess
- Submission queue → Dead-letter queueFailed messages
Technology stack
Frontend
- React 18
Role-specific interfaces, React Router, Redux Toolkit and Axios.
- Monaco Editor
Programming answers and code editing.
Backend
- Spring Boot
REST controllers, DTOs, application services and submission consumer in one application.
- Spring Security
JWT authentication and role-based access.
Databases & persistence
- MySQL
Spring Data JPA: users, roles, structures, exam metadata and attempts.
- MongoDB
Spring Data MongoDB: questions, submitted answers, logs and idempotency records.
Messaging & email
- RabbitMQ
Spring AMQP: submission queue, internal consumer and dead-letter queue.
- SMTP
Email verification and password recovery.
Research / Work in progress
/ 053D Paint Estimation & AI Platform
Academic project · In Progress
LAB NOTES / 05In Progress
Nova.
A work in progress, with a clear direction.
01
EXPERIMENTAL NOW
Nova · wall-condition analysis
02
PLANNED NEXT
3D visualization · camera measurements · complete estimates
Exploring smarter spaces, surfaces and estimates.
MY CONTRIBUTION
I’m developing and testing the initial Nova assistant and preliminary paint quantity calculations as part of my fifth-year engineering project.
- Experimental wall-image analysis for visible defects.
- Initial quantity calculations using dimensions, coverage and coats.
- Planned: 3D rooms, advanced defect detection, camera dimensions and complete cost estimation.
EXPERIMENTAL NOW
AI-assisted wall-condition analysis and initial paint quantity calculation experiments.
Work in Progress
Nova is a prototype, not a final architecture.
Initial wall-condition analysis and paint quantity calculation experiments are underway. Interactive 3D rooms, camera-based measurements, advanced defect detection and complete cost estimation remain planned or under development.
Experimental prototypePlanned / under development
04Technical repertoire
The tools behind
the thinking.
Technologies used across internships and projects. An evolving repertoire, without proficiency scores.
01
Frontend
ReactNext.jsTypeScriptJavaScriptMaterial UI
React interfaces, Next.js pages and server-side facades in AuditFlow.
02
Backend
ASP.NET Core Web APISpring BootPythonJavaFastAPIREST APIsJWT
Spring Boot business services; ASP.NET Core APIs and AI orchestration.
03
Databases
PostgreSQLMySQLMongoDBEntity Framework CoreHibernate
Relational persistence and complementary document storage.
04
AI & Machine Learning
TensorFlow / KerasMediaPipeResNet50BiLSTMMachine LearningComputer Vision
Separate ResNet50 and MediaPipe / BiLSTM experiments in ASL Translator.
05
Development Tools
Git / GitHubPostmanSwaggerVS CodeScrum
Version control, API exploration and collaborative development.
SOFTWARE ENGINEERING & INTELLIGENT SYSTEMS
05The next chapter
Open to a 5–6 month PFE internship · 2027
What if we
worked together?
I am currently seeking a 5–6 month final-year internship starting in February 2027, where I can apply my software development and artificial intelligence skills to practical, innovative projects.
Have an opportunity or a project to share? Let’s talk!