I'm a software developer student at 42 Angoulême, interested in the point where software engineering, infrastructure and security meet.

I like understanding the whole system: the application, its APIs, the Linux host, the network and the controls that keep everything observable and resilient.

Build it. Break it. Understand why. Automate the lesson.

This is how I connect the different parts of a project:

flowchart LR
  Client[Client] --> API[API]
  API --> Backend[Backend]
  Backend --> Data[(Database)]
  Backend --> Queue[(Redis)]
  Backend --> Host[Linux host]
  Host --> Observability[Logs and monitoring]
  Observability --> Security[Security feedback]
  Security --> Backend

  classDef app fill:#164E63,stroke:#67E8F9,color:#ECFEFF,stroke-width:2px
  classDef data fill:#312E81,stroke:#C4B5FD,color:#EDE9FE,stroke-width:2px
  classDef infra fill:#14532D,stroke:#A7F3D0,color:#DCFCE7,stroke-width:2px
  classDef security fill:#713F12,stroke:#FDE68A,color:#FEF3C7,stroke-width:2px

  class Client,API,Backend app
  class Data,Queue data
  class Host,Observability infra
  class Security security

The projects below are representative of the questions I like to explore. Each one starts with a system, a failure mode or an assumption worth testing.

How different is the documented API from the application that actually runs?

An automation tool for comparing an OpenAPI surface with observed application behaviour and highlighting security-relevant gaps.

OpenAPI REST Security Automation

flowchart LR
    Spec[OpenAPI spec] --> Scanner[Scanner]
    Target[Running API] --> Scanner
    Scanner --> Findings[Findings]
    Findings --> Report[Security report]

  classDef source fill:#164E63,stroke:#67E8F9,color:#ECFEFF,stroke-width:2px
  classDef process fill:#14532D,stroke:#A7F3D0,color:#DCFCE7,stroke-width:2px
  classDef result fill:#713F12,stroke:#FDE68A,color:#FEF3C7,stroke-width:2px

  class Spec,Target source
  class Scanner process
  class Findings,Report result

What does automated traffic reveal when a service is deliberately exposed?

A self-hosted environment for collecting, parsing and analysing suspicious activity around exposed services.

Linux Docker Networking Python

flowchart LR
    Internet --> Gateway[Nginx]
    Gateway --> Honeypot[Honeypot service]
    Honeypot --> Logs[Raw logs]
    Logs --> Parser[Parser]
    Parser --> Store[(SQLite)]
    Store --> Dashboard[Analysis]

  classDef external fill:#312E81,stroke:#C4B5FD,color:#EDE9FE,stroke-width:2px
  classDef service fill:#14532D,stroke:#A7F3D0,color:#DCFCE7,stroke-width:2px
  classDef detection fill:#713F12,stroke:#FDE68A,color:#FEF3C7,stroke-width:2px
  classDef analysis fill:#164E63,stroke:#67E8F9,color:#ECFEFF,stroke-width:2px

  class Internet external
  class Gateway,Honeypot service
  class Logs,Parser detection
  class Store,Dashboard analysis

How can personal services stay simple, observable and reasonably protected?

Linux infrastructure for containers, reverse proxying, HTTPS, DNS and security controls.

Linux Docker Nginx Cloudflare

What changes when inference runs locally instead of behind an API?

Experiments with local inference, quantization, context size, memory usage and developer tooling.

Ollama llama.cpp Open WebUI Linux

42 has been my foundation for systems programming, Unix environments, networking and collaborative development.

I'm interested in backend development, security engineering and infrastructure projects where understanding the system matters.

• Website: caillier-romain.fr
• LinkedIn: romain-caillier
• GitHub: rom98759
• Email: rcaillie@student.42angouleme.fr