Search with Jev and Milvus
Milvus bootcamp notebooks combining Gemini embeddings, Milvus retrieval and Jev judgments for reranking, filtering, routing and stopping.
Overview
This entry points to Milvus's bootcamp repository of tutorials and demos. According to the directory summaries, it includes nine runnable Python notebooks that combine Gemini embeddings and Milvus vector retrieval with Jev Noul and Choice judgments, while Python code applies the policies: reranking, filtering, deciding when to stop searching, routing, cache reuse, curation, guardrails and evaluation, on synthetic examples.
The fetched bootcamp README is a general index of Milvus tutorials (RAG, hybrid search, image search, recommendations) and does not mention Jev, so the notebook specifics come from list summaries. The pattern shown is the standard split: vector search provides recall, Jev makes typed judgments about candidates, and code owns the ranking, thresholds and stopping rules.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Is this retrieved passage relevant to the query?(inferred) | — |
| Yes / no | Is there enough evidence to stop searching?(inferred) | — |
| Choice | Which route or collection should this query go to?(inferred) | — |
How it works
- Embed documents with Gemini and index in Milvus
- Retrieve candidates for a query
- Ask Jev Noul/Choice questions about candidates or routes
- Apply ranking, filtering, routing and stopping policies in Python
Strengths
- Runnable notebooks from an established vector DB project
- Covers many search decision patterns
Caveats
- Jev content not visible in the fetched README
- Synthetic examples
At a glance
- Type
- Article
- Status
- production
- Language
- Python
- Runtime
- Jupyter notebooks
- Has code
- yes
- License
- Apache-2.0
Use cases and integrations
How curated lists describe it
Search engineering: nine runnable notebooks combine Gemini embeddings and Milvus retrieval with Jev Noul and Choice judgments, while Python applies ranking, filtering, routing, and stopping policies to synthetic examples.
yibie/awesome-jevNine runnable Python notebooks combining Gemini embeddings, Milvus retrieval, and Jev decisions for reranking, filtering, search stopping, routing, cache reuse, curation, guardrails, and evaluation
AnotiaWang/awesome-decision-modelsRunnable search tutorials using Gemini embeddings, Milvus retrieval, and Jev judgments for reranking, filtering, and routing.
logicrw/awesome-jev-projectsNine Python notebooks combining Gemini embeddings, Milvus retrieval, and typed Jev decisions for reranking, filtering, search stopping, routing, cache reuse, curation, guardrails, and evaluation.
cobanov/awesome-jevNine runnable notebooks combine Gemini embeddings and Milvus retrieval with Jev judgments for reranking, filtering, search stopping, routing, cache reuse, curation, guardrails, and evaluation.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev do in these notebooks?
Per list summaries, it makes Noul and Choice judgments for reranking, filtering, routing, stopping and guardrails.
Which embedding model is used?
Gemini embeddings, stored in Milvus.
Who decides the final ranking?
Python policies apply the thresholds and ordering.
Related projects
- Jev Search★ 534
Plain-language web search where a decision model like Jev picks query, sources and time range, then ranks Search1API results.
AppproductionTypeScript Chrome extension that re-ranks Google results with Jev questions per result and folds away sales pages and SEO filler.
Browser extensionbetaTypeScriptBrowser extension that has Jev score search results for relevance, depth, SEO filler and page type, then re-ranks them with a 0-100 badge.
Browser extensionprototypeJavaScriptResearch agent where GPT plans and Jev finds evidence in parallel across document regions with Choice and Noul.
Browser extensionresearchTypeScriptshadcn/ui site-search block that shows keyword hits instantly, then has Jev re-rank the top 20 by what the visitor meant.
DemobetaTypeScriptDeepSeek Harness paper-reading workbench whose PDF search reranks keyword candidates with Jev, falling back to keyword order.
BenchmarkbetaTypeScript