Camil Benameur, AI engineer.
At Schneider Electric, on an internship then a fixed-term contract (12 months), I designed an installer for industrial software and shipped it to production, and I built internal AI tools there. I am looking for an AI engineer or Forward Deployed Engineer role, building AI tools with the teams who use them.
Lyon, France · graduated in 2025 · available now
Work
Each item shows its status: done at work, public or private. Planned projects are listed separately, further down.
- At work: SoftDPAC Remote InstallerA Go and React application that installs industrial automation software on air-gapped industrial PCs over SSH. Shipped to production.Schneider Electric · April 2025 to March 2026
- At work: Internal AI toolsA custom MCP server, a RAG agent over Confluence documentation and an AI code-review pipeline.Schneider Electric · October 2025 to March 2026
- At work: Internal tools and deploymentGuarded writes to a SQL Server database (tested in a sandbox), applications moved to containers, automated deployment (CI/CD).Call To Action · August to October 2026
- Public: go-llm-streamA Go library that parses an LLM’s JSON while it is still streaming, and repairs it when the answer is cut off.GitHub · MIT licence · v1.3.0
- Private, on request: Public data pipelineJoins the French business register (Sirene API) with a places dataset (Overture Maps), then audits websites while respecting robots.txt. 1,191 tests.Personal project · Oct. 2026
Background
Graduated in 2025. Twelve months at Schneider Electric on one product, from internship to fixed-term contract (April 2025 to March 2026), then about two months at Call To Action (August to October 2026).
- Languages
- French (native), English (C1, TOEIC 910/990), Spanish (elementary).
- Certifications
- Advanced Golang Concepts, Software Design and Architecture Specialization.
- HEAD available now
Data & AI Engineer, permanent role
Call To Action (Bonum Group), Nice
Internal tools, data and automated deployment. The role ended on 16 October 2026.
Software Engineer, fixed-term contract
Schneider Electric, Carros
Owned the SoftDPAC Remote Installer until it shipped to production. Internal AI tools and the Copilot Champion role.
Software Engineer Intern
Schneider Electric, Carros
Designed the same installer from scratch, in Go and React, under strict cybersecurity requirements.
Engineering degree and MSc
Polytech Dijon and Université de Bourgogne
Engineering degree in AI and software development (Polytech Dijon, 2025). MSc in Databases and AI (Université de Bourgogne, 2025).
Lab intern, computer vision
Universiti Teknologi PETRONAS, Malaysia
Embedded stereovision for 3D capture and distance measurement, stereo camera calibration, real-time object detection (YOLOv8).
Skills
Three tables: what I used at work, what my public projects show, and what I have not used yet.
Used at work
- applied_ai
custom MCP server, RAG agent, AI-assisted code review, local assistant (Ollama, Open WebUI), computer vision (YOLOv8, during a lab internship)
- languages
- Go
- Python
- TypeScript
- JavaScript
- SQL
- Bash
- PowerShell
- web_api
- React
- Next.js
- FastAPI
- REST
- SSE
- Tauri 2.0
- tests
unit and integration tests, end-to-end Cypress
- ci_cd
- GitHub Actions
- Docker
- Docker Compose
- Podman
- security
key-based SSH, air-gapped PCs, database writes with a reviewed plan and rollback, GitHub secrets
- data
SQL Server, PostgreSQL, MariaDB, database design (MSc in Databases and AI)
- systems
- Linux (Debian, Ubuntu)
- Windows
- architecture
hexagonal architecture
Visible in my public projects
- llm
parsing streamed LLM JSON, in the event formats of OpenAI and Anthropic (go-llm-stream)
- quality
fuzzing, benchmarks published with their limits
- web
Astro, this site and my blog, deployed on Cloudflare
Not used yet
- agents
- LangGraph
- llm_apis
- OpenAI
- Gemini
- evaluation
agent test sets, automatic grader compared with a person
- tracing
OpenTelemetry, Langfuse, cost per answer
- cloud
- AWS
- Google Cloud
- Azure
- Terraform
- Kubernetes
No professional experience yet with AWS, Google Cloud or Azure. I have used Anthropic’s API on personal projects, not yet OpenAI’s or Google’s. The projects coming next aim to fill some of these gaps.
Coming next
The projects I plan to build next, in public. None of them exists yet: a box gets ticked, with a link, when the project is online.
- A LangGraph agent with human approval, tested by real users, and a framework-free version to compare. (Planned)
- A public evaluation bench: 100 hand-labelled cases, an automatic grader compared with a person, error margins. (Planned)
- The agent’s trace log, with its cost per accepted answer. (Planned)
- The same test set run on Anthropic, OpenAI, Gemini and a local model. (Planned)
- A Google Cloud Run deployment described as code (Terraform), with federated identity, a budget and a kill switch. (Planned)
Writing
I write about what I built, and I flag at the top of a post what I corrected.
- How an Offline Installer Works, and What I Over-Built Along the WayFrom SSH deployment to the offline package relay, and what I would change today.
- From Backlog to MVP: Building an Offline Industrial Installer in GoInstalling Docker offline from .deb packages, with version 27.5.1 as the default.corrected on
- Streaming LLM JSON in Go: a Resumable Scanner and a HealerReading a model’s answer while it arrives, and repairing an answer that was cut off.corrected on
Contact
A role, a question or a comment on a project: write to me, I answer in writing.
I am also preparing a freelance AI engineering activity. It does not exist yet.