Research notes
Reading the geometry of AS-OCT pretraining
Measured representation changes across five AS-OCT pretraining runs, and the profiling work that cut two-GPU JEPA update time by 14.2%.
Nassim
Louissi
Research Engineer
I build and operate ophthalmic research systems, from hospital data pipelines to machine learning experiments.
At Quinze-Vingts Hospital in Paris, I built CorneaForge and its OphtaFlow AI interface. I also study how models learn from corneal images and engineer the training systems that let me test those ideas.
Corneal geometrySchematic
Geometry becomes a research question.
Production systems
Device ingestion, image processing, annotation, and model results in an on-premise hospital platform.
Research experiments
Five pretraining recipes on 3.43 million AS-OCT images, plus Corvis motion analysis and neural video classification.
Measured performance
Two-GPU JEPA updates reduced from 10.246 to 8.787 seconds, with the training recipe and resume behavior preserved.
CorneaForge / OphtaFlow AI
A hospital platform I built and operate.
CorneaForge connects ophthalmic devices to a working research environment. Its deployed services ingest MS-39 and Corvis data, process OCT images, compute corneal features, and render examinations for the OphtaFlow AI interface.
The application brings examinations, native measurements, videos, annotation, and existing model predictions into one place. Retained source objects connect the research data back to the device exports.
Read the architecture and engineering decisions- Data
- MinIO for source objects; PostgreSQL for structured measurements and processing state.
- Processing
- Dedicated ingestion, feature, rendering, and OCT workers supervised by systemd.
- Application
- FastAPI and OphtaFlow AI connect examination review, annotation, and model outputs.
Writing from the work
The questions, decisions, and implementation details behind my research.
Representation learning
Learning without labels: what MAE and JEPA actually predict
Follow one image from pixels to representations, calculate a training loss, and see why agreement alone can teach a model nothing useful.
Experimental design
Five OCT pretraining recipes, and what a fair comparison means
The five AS-OCT recipes I implemented on 3.43 million images: their inputs, objectives, shared controls, and compute tradeoffs.
Research systems
From a device export to a research system
How I built and operate CorneaForge / OphtaFlow AI: device ingestion, scientific processing, examination review, and model serving on hospital infrastructure.
Scientific computing
From measured points to Zernike coefficients
How I turn native corneal samples into maps and Zernike descriptors in CorneaForge, with explicit reference surfaces, coordinates and units.
Performance engineering
Cache the geometry, then change the values
How I implemented reusable interpolation operators in CorneaForge, including missing-data cache keys and a measured memory budget.
Open source
Tools and experiments that grew out of questions I needed to answer.
Monitoring & profiling
ML Training Monitor
The monitoring and profiling workflow I use in my own research, packaged as a skill for coding agents.
I used it to trace coordinator stalls in two-GPU JEPA, overlap input preparation with GPU work, and measure a 14.2% reduction in update time. Read the profiling result.
View the repository- Observe the run
- Bring training metrics and GPU activity into the same view.
- Find the bottleneck
- Use profiles to investigate where time and memory go.
- Check the change
- Compare measurements and numerical behavior after an optimization.
OCT-CUDA
A CUDA experiment in fusing an OCT reconstruction pipeline with cuFFTDx, exploring the cost of moving intermediate values through GPU memory.
Selected publications
Machine Learning Model for Predicting Visual Acuity Improvement After Intrastromal Corneal Ring Surgery in Patients With Keratoconus
Perez E*, Louissi N*, et al.
CorvisST biomechanical indices in the diagnosis of corneal stromal and endothelial disorders: an artificial intelligence-based comparative study
Borderie VM, Georgeon C, Louissi N, et al.
Have a research question in common?
I’m happy to talk about ophthalmic ML, scientific computing, and the practical work behind an experiment.
louissi.nassim@gmail.com