A consultancy operating in the space where computer vision, machine learning, and applied AI meet the problems they were built to solve.
About
I'm J.J. (Jean-Jacques) Larrea, an EE/CS graduate from Princeton with a focus on computer graphics and image processing. My career has been spent at the pixel level: production systems for computer animation and visual effects, automated film scan/record/calibration, digital compositing, color separation, and colorspace transformation.
Two shifts changed what a solo practitioner can credibly take on: AI Vision and Vision-Language models have matured into reliable building blocks; AI-assisted development has collapsed the team-sized engineering effort previously required to ship ambitious systems. Op Null is what I'm doing about it.
Engagements
Under Op Null I'm building a number of systems for problems where the inputs are heterogeneous, unlabeled or partially labeled, and the output facts need to be defensible. Recurring techniques in the work:
- Visual embeddings and vector search for cross-session, cross-source object identity
- Spatial and temporal coherence as anchors across noisy or inconsistent capture
- Confidence-tiered human-in-the-loop curation workflows
- Hybrid CV / vision-language-model inference — fast triage with deep analysis on demand
- Progressive, corpus-specific model improvement from reviewer actions
Two provisional patents are filed; more are in progress. Proof-of-concept applications are at varying stages.
Get in touch if you'd like to:
- Collaborate on a project, mine or yours
- Bring a problem in applied vision you think is worth working on