OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

encoder-run

encoder-run is a Kubernetes operator that deploys embedding models and supporting storage, monitors source-code repositories, and refreshes vector embeddings as code changes. Its deployed gateway and console UI let users search and use those embeddings for semantic code search, RAG, similarity checks and code analysis.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

encoder-run — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use encoder-run when you need to operate a self-hosted, Kubernetes-based pipeline that indexes source code and refreshes embeddings as repositories change.

Who it is for

  • Teams operating Kubernetes-based semantic search over source-code repositories.
  • Developers building code-focused RAG or code-analysis workflows that need synchronized embeddings.

Documented uses

  • Deploy configurable SentenceTransformer or encoding models for source-code embedding.
  • Keep repository embeddings current as code changes.
  • Use generated vectors as a retrieval layer for semantic search, similarity checks, RAG and code analysis.
Before you choose

Requirements & limits

The project is a public GitHub repository under the Apache-2.0 license.

  • Embedding dimensions are hard-coded to 768, so the chosen transformer must match.
  • The README recommends at least 24 GB of memory for the Docker Desktop setup and notes that model weights and images also require reasonable disk capacity.
  • GPU-backed inference can speed up embedding compared with local inference.
  • Node 18.17.0 is listed only for local frontend development.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Deploy encoder-run to a local Kind cluster

Set up the documented local Kubernetes deployment and open its console and gateway endpoints.

Before you start

  • Go 1.21 or later
  • Docker
  • Kubernetes Kind, installed separately or through make kind-install
  • At least 24 GB of memory recommended for Docker Desktop
  • A transformer model with 768-dimensional embeddings

Steps

  1. Ensure Docker is installed and available.

  2. Install Kind, using make kind-install if needed.

  3. From the project setup, run make deploy to build images from source, or make deploy-staging to pull images from Docker Hub.

  4. After deployment, access the console UI at localhost:32081 and the gateway at localhost:32080.

Expected outcome

The documented components are launched in Kubernetes, with the console UI and gateway exposed on the listed local ports.

Guide limitations

  • The first source build can take more than 10 minutes.
  • The setup recommends substantial memory and disk capacity for model weights and container images.
  • Local embedding is slower than GPU-backed inference.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
encoder-run/operator · Official homepage title · retained in its source language
How we review sourcesSuggest a correction