LCLlewellyn Christian
Back to Projects
Computer vision + evidencePublic beta
Flagship project

ODE Vision Evidence Lab

A public computer-vision interface and API surface designed to pair visual results with model context, documentation, and honest release boundaries.

Vision systems earn trust by showing what ran, what the system returned, and which claims still require controlled benchmarking.

Live public interfaceDeveloper documentationPublished limitations
What this demonstrates
The public vision interface is reachable without a payment step.
Live OpenAPI documentation exposes the currently supported developer surface.
Free community access is separated from private deployment and support.
Selected stack
PyTorchRT-DETRFastAPIOpenAPIComputer visionModel provenance
Published limitations
  • No public benchmark report is claimed on this page yet.
  • The separate 3D training notebook remains in development pending real-dataset validation.
  • Community availability is fair-use and does not include an enterprise service-level agreement.
Experience mode
Step 1
Image
Step 2
Detector
Step 3
Objects
Step 4
Review
Visible inference
A result is more useful when the operator can inspect how it was produced.
Live pattern
Engineering lens
  • The interface makes the input and output path visible.
  • Labels and confidence belong beside the result, not hidden in a demo script.
  • Human review remains part of consequential use.
Platform fit
This project belongs in Llewellyn Systems because it turns a repeated engineering pattern into a governed operating asset. The page is a public architecture surface; executable evidence is linked separately when available.
Publication boundary
Stack names describe the public abstraction layer. Employer information, customer data, secrets, and proprietary implementation details stay outside this showcase.
Related projects
Skills + MCP

Agent Fabric

Engineering prototype
Skills + MCP
Agent Fabric
Prototype
IntentSkill loaderTool contractResult

A governed capability-routing pattern that exposes named MCP tools, matches work to specialists, and applies policy and audit boundaries before execution.

PythonMCPJSON SchemaCapability registry

A governed capability-routing pattern that exposes named MCP tools, matches work to specialists, and applies policy and audit boundaries before execution.

Capability routing
Tool scoping
Human review gates
PythonMCPJSON Schema
Open
JupyterBook + MyST

Notebook Loom

Engineering prototype
JupyterBook + MyST
Notebook Loom
Prototype
OutlineMySTSectionsBook

A reproducible publication pattern that keeps MyST narrative, executable notebooks, and public technical guides in version-controlled source.

MySTJupyterBookJupyterMarkdown

A reproducible publication pattern that keeps MyST narrative, executable notebooks, and public technical guides in version-controlled source.

Executable docs
Live widgets
Published learning assets
MySTJupyterBookJupyter
Open
Data + AI governance

Governance Spine

Engineering prototype
Data + AI governance
Governance Spine
Prototype
AssetOwnerPurposeRegistry

An executable governance pattern for classifying risk, blocking unsafe changes, recording decisions, and checking public routes before release.

Policy as codePythonApproval gatesAudit logs

An executable governance pattern for classifying risk, blocking unsafe changes, recording decisions, and checking public routes before release.

Lineage
Approvals
Rollback
Policy as codePythonApproval gates
Open
ODE Vision Evidence Lab Public beta | Llewellyn Christian