Add on-device eval agent example (ZEDCLOUD-2462)#31
Open
adithya-zededa wants to merge 1 commit into
Open
Conversation
Adds the ondevice-eval-agent application under edgeai/ as an example that runs alongside an inference server (Triton / OpenVINO) at the edge. It provides: - Flask backend that proxies inference and exposes model discovery / introspection tools via an MCP-style tool registry - React SPA for chat-driven model exploration, inference, and result visualization - LLM router that supports Anthropic, OpenAI, Google, Groq, Ollama, and any OpenAI-compatible endpoint, with optional rate-limit and retry handling - Single-image Dockerfile that builds the SPA and serves it together with the API on port 8080 Reads all API keys from environment variables; no secrets are bundled.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Opens the ondevice-eval-agent application as an example under
edgeai/ondevice-eval-agent/. The agentruns alongside an inference server (Triton / OpenVINO) at the edge and
exposes:
surfaces ~16 MCP-style tools the LLM can call to introspect tensors,
generate integration code, and run inference end-to-end.
visualization of inference results.
Ollama, and any OpenAI-compatible server such as vLLM / LM Studio /
TGI) with rate-limit, retry, and SSE streaming support.
serves it together with the API on port 8080.
All credentials are read from environment variables — no secrets are
bundled. Internal-only docs (
CODE_REVIEW_REPORT.md,DEBUG-REQUIREMENT.md,FEATURES_PRESENTATION.md), build artifacts(
node_modules/,dist/,__pycache__/,.pytest_cache/,tsconfig.tsbuildinfo,package-lock.json), and internal Jira ticketreferences in code comments have been scrubbed before this PR.
Jira: ZEDCLOUD-2462
Test plan
cd edgeai/ondevice-eval-agent && docker build -t ondevice-eval-agent .succeeds.docker run --rm -p 8080:8080 -e MODEL_SERVER_URL=http://triton:8000 -e ANTHROPIC_API_KEY=... ondevice-eval-agentstarts andGET /agent/statusreturns 200.http://localhost:8080/and the chat panel reaches the backend.pip install -r requirements.txt && pytest tests/.