#5353rd of 65 in Model servingFor finance and business, e-commerce, marketing and sales
metarank
Ranking service that reranks search and recommendation results from click signals
- Stars
- 2.5k
- License
- Apache-2.0
- Last commit
- Oct 2026
- Last release
- Sep 2026
- Language
- Scala
2,401 on 21 Apr 20262,459 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Metarank is a Scala service that reranks items from an existing search or recommendation system using LambdaMART models trained on visitor events such as clicks and purchases. It exposes HTTP endpoints for ranking and feedback, computes built-in ranking features (CTR, referer, User-Agent, session), and supports trending and similar-item recommendations via matrix factorization (ALS). State lives in Redis for horizontal scaling; a standalone mode trains and serves in one process.
Who it is for: Engineers adding personalized reranking to e-commerce search or recommendations
Strengths
- Reranks large result sets in 10-20ms, per the project's benchmarks
- Dozens of built-in ranking features, so little custom feature code
- Standalone mode trains and serves in one command with in-memory store
- Stateless service with Redis state, scales horizontally on Kubernetes
Weaknesses
- Needs a click and impression event stream to train useful models
- README leaves semantic search and recommendation guides as TODO placeholders
- Reranks results only; you still need a separate retrieval engine
- Not a drop-in for LLM or RAG pipelines despite the LLM encoder features
What it needs
- Needs Redis (production state store), Elasticsearch/OpenSearch (optional, for LTR integration)
- Models: LambdaMART (XGBoost), Matrix factorization ALS, LLM bi-encoder and cross-encoder text features
- port 8080
Quick answers
Is metarank open source?
Yes. Its code is on GitHub under the Apache-2.0 license.
Can you run metarank with Docker?
Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.
Also in Model serving
See all 65| Rank | Project | Score |
|---|---|---|
| 1 | llama.cppC/C++ inference engine for GGUF LLMs and VLMs with an OpenAI-compatible server | 94 out of 100 |
| 2 | OllamaLocal model runner with CLI and REST API for open-weight LLMs | 92 out of 100 |
| 3 | LocalAISelf-hosted inference server with OpenAI-compatible API and pluggable backends | 91 out of 100 |
| #4 | vLLMInference engine serving LLMs with PagedAttention and an OpenAI-compatible API | 88 out of 100 |
| #5 | strataRuns a 125B-parameter MoE model on one consumer GPU plus system RAM | 88 out of 100 |