Hi, I’m Olivier!
I’m a developer who also takes care of the infrastructure. These days that’s for a genomics research programme at Université Laval; at home it’s a homelab that’s grown a bit out of hand.
Research
Research data platforms
At Genovalia, I look after where the research data lives: the data catalogue, the schemas and metadata standards, the API other tools use, and the OpenShift cluster it all runs on.
- writing software since
- 2010
- years building research software
- 8+
- running production on OpenShift since
- 2022
- working languages
- FR · EN
Dev
Web apps, all the way to production
I build web apps from the API to the page, and I also handle what comes after: containers, CI/CD, identity and access. I like staying with a project once it’s live.
- Push
- Tests and accessibility
- Build and roll out
- Smoke check
Lab & hobbies
The homelab
A few machines at home where I try things out: a local LLM with a gateway in front of it, my notes as a knowledge graph I can search, and MCP servers so AI assistants can use both.
- Docker Swarm
- Caddy
- GitHub Actions
- Gatus
- Local LLMs
- MCP
Research infrastructure
All projects-
Genovalia data platform
A catalogue of research datasets, standardised description schemas, a metadata API and a data explorer, running on Université Laval’s OpenShift cluster. Designed, built and operated end to end.
-
Metadata standards for animal genomics
DCAT-based dataset descriptions in JSON-LD, with controlled vocabularies, ontology mappings, OCA schemas and automated validation, so public genotyping datasets are curated once and reused across projects.
-
Research platform identity
Single sign-on for Genovalia’s applications with Keycloak, OpenID Connect and ORCID, and roles for the internal team and external partners.
-
Automated content analysis
Machine learning pipelines over large corpora of political and media text for quantitative social science research at the CLESSN, evaluated against human-coded samples.
-
Survey and salience-analysis platform
One tool for designing survey instruments, collecting responses and analysing issue salience, used by CLESSN researchers and students.
-
Applied ML and LLM tooling
Machine learning on large text corpora, checked against human coding; and now local LLMs, MCP servers, and the rules for what they’re allowed to do.
Software and operations
All projects-
Websites and web apps
Full stack, from the API to the page. FastAPI and Django on the back end; SvelteKit, Vue and Nuxt on the front end; Wagtail for content sites. This site is one of them.
-
Containers and hosting
OpenShift and Kubernetes in production, Docker Swarm and Caddy for smaller setups. I set them up, watch them, upgrade them, back them up, and fix them when something breaks.
-
CI/CD and automation
GitHub Actions, with my own runners where it makes sense, and Kustomize manifests. Every push is tested, a push to main deploys, a bad deploy rolls back on its own, and dependencies stay up to date.
-
Identity, access and Microsoft 365
Sign-in for platforms with Keycloak, OpenID Connect and ORCID, with roles for each team and partner; and Microsoft 365 and Exchange: accounts, administration and migrations.
-
Projet Quorum
The Django backend of a civic participation platform.
-
This website
Plain HTML built ahead of time with SvelteKit, in two languages. A strict security policy (no outside scripts), preview images and a PDF resume made at build time, and accessibility checks in its tests.
At home
All projects-
Personal knowledge graph
My notes, people, places, projects and tasks as a typed graph with semantic search, and an MCP server so an AI assistant can file things for me.
-
Tasks for a local LLM
A small service in front of the LLM running on my own GPU: other apps call a named task and get back a decision, a JSON document or text, without knowing about prompts or models.
-
Homelab automations
My home automations as small Python scripts: an AI can write them, but I approve each one, and they run in a sandbox that can only do what they declared.
-
Event log
A timeline of things that happened: places, purchases, trips, deploys. My apps and AI assistants write to it, and I can ask it what we did last spring.
-
Browser home page
The page my browser opens on: a clock and the weather over a drawn scene that follows the time of day, my links, and cards from my calendar, email and notes.
-
Groceries and meal plans
A grocery list that sorts itself by store aisle, and a weekly meal plan with our recipes. The list lives in Home Assistant, so anyone at home can add to it.
-
The cluster
Docker Swarm behind a Caddy gateway, a self-hosted CI runner, health checks, and push-to-deploy for every app, this site included.
Where I’ve worked
- Genovalia
- CRDM
- Université Laval
- CLESSN
- Edgenda
- Fujitsu
- ACPE inc.
- Frima Studio
Let’s talk
Working on a research data platform, or sorting out who gets access to what across teams? Write me, I’m always happy to talk about it!