Design, development and deployment of an AI-assisted career platform (GLYBETTE) using a microservices architecture

Presented by

Rayen Attia

Academic Supervisor Dr. Raik Aissaoui
Host Organization Update IT & Training
Enterprise Supervisor Mr. Aymen Dardouri
University Year 2025/2026

Presentation Plan

Agenda

1

Project Context

2

Market Analysis

3

Vision & Objectives

4

Ecosystem

5

Stack & Architecture

6

Project Realization

7

Conclusion & Retrospective

1

Project Context

1 · Project Context

Problems We Aim to Solve

#1

Scattered Opportunities

The job market in Tunisia is fragmented: offers are scattered across dozens of platforms with no unified view.

2. Integration

Difficult Integration

Graduates struggle to find PFE topics, internships, or jobs matching their profile.

3. Fragmentation

Isolated Career Tools

No single solution covers search, CV, interviews, and training end-to-end.

4. Visibility

Recruiters Lacking Visibility

Companies struggle to identify the best candidates among applicants.

1 · Project Context

Host Organization

Update IT and Training logo Host Company
ManagerMr. Aymen Dardouri
SectorIT & Services
DomainSoftware · Salesforce · AI
SizeSmall & Innovative Start-up
Websiteupdate-it.fr
Productivity User needs
Glybette logo Entrepreneurship Space

Dedicated space for students to transform ideas into concrete projects.

Digital incubator with mentoring, resources, and support throughout the journey.

Incubation Innovation Support

2

Market Analysis

2 · Market Analysis

Competitive Landscape

Resume.io screenshot

Resume.io

CV builder & templates. Strong on documents, weak on aggregation and matching.

Hi Interns screenshot

Hi Interns

Internship-focused in Tunisia. Limited scope beyond listings.

LinkedIn screenshot

LinkedIn

Global network. Broad reach, not tailored to student paths in Tunisia.

Unified ecosystem for fostering young talent; existing tools address isolated parts of the career journey.

2 · Market Analysis

Comparative Feature Analysis

Feature Glybette Resume.io Hi Interns LinkedIn
CV BuilderNoNo
Portfolio BuilderNoNoPartial
Job AggregationNo
AI MatchingNoNoPartial
Interview PrepNoNoNo
Project CollaborationNoNoNo
Market StatisticsNoNoPartial
Tunisia-focusedNoNo

3

Vision & Objectives

3 · Vision & Objectives

Our Vision & Objectives

One place to build, prepare, find and innovate.

Centralizing Offers

Aggregate internship, PFE and job offers from the best sources in one place.

1
Uniting Innovation

Collaborate and propose project topics across disciplines.

3
Informing Decisions

Market stats, salaries, and in-demand skills at a glance.

5
Personalizing Careers

AI matching tailored to each user's field and goals.

2
Serving Businesses

Recruiters post offers and rank candidates efficiently.

4
Supporting Studies

Training, certifications, and school choice across all fields.

6

4

Ecosystem

4 · Ecosystem

Glybette's Ecosystem: Students & Candidates

Students / Candidates
  • CV Builder
  • Portfolio Builder
  • Cover Letter Enhancer
  • Mock Interviews
  • Jobs Repository
  • Offers Matchmaking
  • Collaboration Platform
  • Statistics & Insights

4 · Career Tools

CV Builder & Cover Letter Enhancer

1

CV Builder

  • Professional CV in minutes
  • Domain-specific templates
  • ATS-optimized metrics
  • PDF export & online sharing
  • Compatibility score with offers
  • AI-powered enhancements
2

Cover Letter Enhancer

  • Auto-adjustment per selected job
  • Profile and skills integration
  • Sector and company adaptation
  • Multi-language support
  • AI-powered enhancements

4 · Career Tools

Portfolio Builder

1

Projects Showcase

Description, tech stack, etc.

2

Accessible Page

name.glybette.com, sent directly to recruiters.

3

Domain Templates

IT, Design, Finance, Medicine, Marketing…

4

Certifications

Coursera, AWS badges with verification.

5

Experience Timeline

Visual academic and professional journey.

4 · Offers & Matching

Offers & Intelligent Matching

Sources
IndeedLinkedInWTTJPFE Books

Aggregated from multiple job boards and university listings.

Matching Engine
  • Profile ↔ offer compatibility score
  • Daily personalized recommendations
  • CV & cover letter enhanced per offer
  • Email notifications & tracked applications
Offer Types
InternshipPFEFull-timeRemoteFreelance

AI-powered pipeline from scrape to recommendation.

4 · Interview Prep

Interview Preparation

Job Analysis

AI generates tailored questions for the field and company.

1

Simulation

Interactive Q&A: AI evaluates clarity and structure.

2

Feedback

Gap detection, reformulation suggestions, overall score.

3

4 · Collaboration

Projects Collaboration Platform

1

Post an Idea

Publish a PFE topic with description, field, and skills.

2

Join a Project

Browse ideas and request to join based on skills.

3

Collaborate

Shared workspace, task tracking, messaging.

4

Launch & Promote

Projects showcased in student portfolios.

5

Stack & Architecture

5 · Architecture

Why Microservices?

General benefits of splitting a monolith into services

Monolith vs Microservices

Scalability

  • Scale only the services under heavy load
  • Add capacity where demand grows, not across the whole app

Independent Deployment

  • Release, patch, or roll back one service at a time
  • Each service can use the language, runtime, and infrastructure best suited to its workload

Loose Coupling

  • Clear boundaries between business capabilities
  • Failures and refactors stay contained within a service

Database per Service

  • Each microservice owns its private data store
  • Teams evolve schemas without cross-service table coupling

05 · Architecture

Microservices Architecture

React Client API Gateway Users CV Portfolio Jobs Projects Scraping Notifications AI Service Internal only RabbitMQ HTTP internal HTTP internal
MongoDB RabbitMQ Better Auth Resend Ollama Internal HTTP

Portfolio ↔ Users (internal resolve) · Jobs ↔ AI · CV ↔ AI · Scraping → RabbitMQ ← Jobs · Notifications

05 · Stack

Full Technology Stack

Glybette global runtime architecture: frontend, API gateway, microservices, RabbitMQ, MongoDB, and technology stack

05 · Deployment

Deployment Tooling

Why Glybette standardizes on containers and an automated pipeline

Docker

One reproducible runtime for every microservice on the VPS.

  • Compose orchestrates the full Glybette stack: gateway, services, MongoDB, RabbitMQ, Redis
  • Same image boundaries in dev and production; each sprint ships an isolated service build
  • Environment injected at deploy time; no secrets baked into images

Jenkins

Automated CI/CD on the VPS, from Git push to live rollout.

  • Pipeline triggered on main: checkout, env injection, preflight, deploy, smoke tests
  • Production .env stored in Jenkins credentials, never committed to Git
  • Fits the Scrum cadence: every sprint ends with a verifiable, repeatable deployment

5 · Deployment

CI / CD Pipeline

From Git to VPS: continuous deployment of microservices

Git Push Monorepo source
Jenkins Build + tests
Docker Build 9 service images
Docker Push To Docker Hub
Docker Hub Image registry
VPS Ubuntu · Docker Engine · Jenkins · Nginx
Docker Pull From Docker Hub
Compose Up Orchestration
Nginx Proxy + TLS
Production Glybette live 24/7

6

Project Realization

6 · Methodology

Why Scrum?

3 releases 9 sprints 1 service / sprint
Plan
Build
Review
Retro

Each sprint cycles through ceremonies, then ships one microservice

1

Fit for Microservices

  • Each sprint = one microservice end-to-end
  • Services map naturally to sprint boundaries
  • Incremental integration, no big-bang deploy
2

Sprint Cadence

  • 3 releases × 3 sprints = 9 total
  • Plan → Build → Review → Retro
  • Continuous supervisor feedback
3

Feedback Loop

  • Weekly demos to enterprise & academic supervisors
  • Backlog refinement between sprints
  • Retrospectives drive process improvements

6 · Releases

Releases & Sprints

Release 1

Platform Foundation

Sprint 1

API Gateway

Sprint 2

Users Service

Sprint 3

Notifications Service

Release 2

Professional Identity

Sprint 4

Portfolio Service

Sprint 5

CV Service

Sprint 6

Projects Service

Release 3

Opportunities & AI

Sprint 7

Scraping Service

Sprint 8

Jobs Service

Sprint 9

AI Service

Release 1 · Platform Foundation
Sprint 1

API Gateway

API Gateway route map
Authenticated proxy sequence diagram
Release 1 · Platform Foundation
Sprint 2

Users Service

Users Service use case diagram
Release 1 · Platform Foundation
Sprint 2

Users Service

OAuth login page
Email OTP verification page
Release 1 · Platform Foundation
Sprint 3

Notifications Service

Notifications Service use case diagram
Notifications Service component diagram
Release 2 · Professional Identity
Sprint 4

Portfolio Service

Portfolio Service use case diagram
Release 2 · Professional Identity
Sprint 4

Portfolio Service

Portfolio builder editor
Release 2 · Professional Identity
Sprint 5

CV Service

CV Service use case diagram
Release 2 · Professional Identity
Sprint 5

CV Service

CV section editor
Release 2 · Professional Identity
Sprint 6

Projects Service

Projects Service use case diagram
Release 2 · Professional Identity
Sprint 6

Projects Service

Project detail page with join-request panel
Release 3 · Opportunities & AI
Sprint 7

Scraping Service

Scraping Service use case diagram
Scheduled scraping activity diagram
Release 3 · Opportunities & AI
Sprint 8

Jobs Service

Jobs Service use case diagram
Release 3 · Opportunities & AI
Sprint 8

Jobs Service

Offer detail page with matching and tabs
Release 3 · Opportunities & AI
Sprint 9

AI Service

AI Service use case diagram
AI Service technical architecture

7

Conclusion & Retrospective

07 · Conclusion

Project Conclusion

Phase 1 · Thesis scope

Phase 1 accomplished

The candidate-facing platform is live: identity, portfolios, CVs, collaboration projects, scraped opportunities, and AI-assisted career tools, delivered across three releases.

9 microservices shipped 3 releases delivered 9 Scrum sprints

Candidate platform

Students and graduates can sign in, build a professional presence, and explore opportunities in one product.

Three releases shipped

Platform foundation, professional identity, and opportunities & AI, each as independently deployable services.

AI in core flows

CV enhancement, offer matching, mock interviews, cover letters, and job chat through internal AI Service.

Async foundations

RabbitMQ handles offer ingestion, notifications, OTP delivery, and profile projections.

Repeatable delivery

Docker Compose per sprint with Jenkins CI/CD: one service deployed end-to-end each iteration.

Ready for Phase 2

Architecture leaves room for recruiters, premium tiers, and institutional partnerships next.

07 · Retrospective

What's Next

Phase 1 is shipped. Here's what comes next.

Robust Infrastructure

  • Harden CI/CD for reliable production rollouts
  • Upgrade VPS capacity for scraping & AI workloads
  • Move off local Ollama to hosted inference for faster responses

Recruiter Features

  • Enterprise offer posting & pipelines
  • Candidate ranking dashboards
  • Recruiter analytics & shortlists

Premium Plans

  • Advanced AI matching & interview prep
  • Priority alerts & enhanced themes
  • Premium portfolio subdomains

Partnerships

  • Universities & engineering schools
  • Incubators & career offices
  • Company onboarding programs

Thank You For Your Attention!