docs: comprehensive UX/UI improvements to homepage, navigation, and content#658
docs: comprehensive UX/UI improvements to homepage, navigation, and content#658alexagriffith wants to merge 12 commits intokserve:mainfrom
Conversation
✅ Deploy Preview for elastic-nobel-0aef7a ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
b6a345d to
08cc5a3
Compare
…ontent - Restructure navbar: standardize to "Getting Started" (was "Get Started"), reorganize Docs dropdown with grouped sections, merge Blog into Community - Rewrite homepage with improved hero, benefits, architecture, adopters, quickstart, and showcase sections; fix broken homepage links - Add HomepageSectionNav component for improved navigation - Create deploy-your-first.md decision guide with comparison table replacing broken generated-index description - Standardize tutorial page titles to "Deploy Your First LLM/Predictive..." naming scheme with (Standard)/(Advanced) clarity labels - Add colorblind-safe syntax highlighting in light mode (orange-brown strings, dark teal numbers, dark blue keys/attributes) - Add "Which Installation Do I Need?" decision table to install/overview.md - Update control-plane-llmisvc.md sidebar label and add purpose/audience intro - Add InferenceService vs LLMInferenceService comparison table to control-plane.md - Update versioned sidebars (0.16, 0.17) to match new Deploy Your First structure - Fix footer label: "Get Started" → "Getting Started" - Update Next Steps links across quickstart-guide.md and llmisvc-install.md to use new consistent page titles Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
064fa03 to
b9d6f55
Compare
|
Question - do we want to say LLM or generative, I changed the getting started to be consistent saying LLM, but I know there is nuance there...so we can change it. |
… scope InferenceService covers predictive and standard LLM workloads, not just 'Standard & GenAI' which was ambiguous. LLMInferenceService is for advanced LLM-only features. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
- GPU Acceleration → multi-node guide (not generic admin overview) - Autoscaling (GenAI) → generative inference autoscaling page - Intelligent Routing → custom transformer docs (predictor/transformer/explainer routing) - Cost Efficient → serverless/scale-to-zero guide Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
Replace wall-of-prose with decision table, workload comparison, and structured best practices sections. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
…ll content Add decision tables, section dividers, and structured headings without removing any existing content. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
…ction Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
These URLs resolve incorrectly in Docusaurus static builds. Replaced: - multi-node/multi-node -> llmisvc-overview (covers GPU/multi-node) - autoscaling/autoscaling -> llmisvc-configuration (covers autoscaling config) - custom-transformer/custom-transformer -> inferencegraph/overview (routing) - serverless/serverless -> kpa-autoscaler (scale-to-zero) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
- GPU Acceleration -> LLMIsvc deployment guide (covers multi-node GPU orchestration) - Autoscaling (GenAI) -> KEDA autoscaler docs - Intelligent Routing -> control-plane (predictor/transformer/explainer architecture) - Cost Efficient -> KPA autoscaler (Knative scale-to-zero, unchanged) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Signed-off-by: Alexa Griffith <agriffith96@gmail.com>
There was a problem hiding this comment.
The home page lacks details about llminferenceservice. hope we can add that.
There was a problem hiding this comment.
I think the getting started is where that lives, the homepage has gen ai usage info
| } | ||
|
|
||
| [data-theme='dark'] .navbar__title { | ||
| color: white !important; |
There was a problem hiding this comment.
Can we try not to use !importent if possible ?
| #60a5fa 70%, | ||
| #93c5fd 100% | ||
| ); | ||
| background: |
There was a problem hiding this comment.
btw, there is inconsistency between dark and light theme.
There was a problem hiding this comment.
the button was what i was trying to fix
There was a problem hiding this comment.
which inconsistency do you mean exactly? i was trying to fix the dark theme, but the buttons just kept staying the same, im not a FE so I may just be missing a style expectancy
There was a problem hiding this comment.
And i changed the style like that intentionally, its a preference of style i guess, i like the new one but if we like the previous banner which I assume you mean the drop down naming? I moved community info to the left, and version of the website to the right as that seems a little less important to the docs than the community docs. let me know exactly what you think should change
|
Not sure if we should remove modelmesh docs as it is archived. |
where do you see it archived? I see it in docs/admin-guide/modelmesh.md |
terrytangyuan
left a comment
There was a problem hiding this comment.
It might be useful to take some screenshots of the improvements to help review
Yes, we can remove modelmesh. I just archived all relevant repos yesterday. |


Summary
This PR overhauls the documentation site's UX/UI across the homepage, navigation, sidebar, and key content pages. No existing content was removed — all changes are additive improvements (better layout, clearer naming, added links, improved visual hierarchy).
While updating the community page I noticed quite a few stylistic and naming inconsistencies and misalignments. so I kind of accidentally went down this rabbit hole of trying to make it all more consistent.
Changes by Area
Navigation & Sidebar
Serve an LLM,Serve an LLM (Advanced),Serve a Predictive Model)InferenceServicevsLLMInferenceServicedistinction in the sidebarlinkproperties to Architecture, Generative Inference, and other categories so breadcrumb navigation clicks through correctlyHomepage
HomepageSectionNavcomponent for smooth scroll-to-section navigationWelcome to KServe (
intro.mdx)intro.md→intro.mdxwith full MDX layout: responsive grid cards, tabbed Key Benefits, tabbed Supported Frameworks grid with doc-card linksTabItembase-path resolution bugGlobal CSS / Syntax Highlighting
doc-grid-2,doc-grid-3,doc-cardresponsive utility classes for MDX pagesContent Improvements
install/overview.md: added "Which Installation Do I Need?" decision table at the top with three paths (KServe only → Predictive; KServe + LLMIsvc → GenAI; Full stack → caching)control-plane.md: addedInferenceServicevsLLMInferenceServicecomparison tablecontrol-plane-llmisvc.md: added intro paragraph explaining why this CRD exists and who should read itgenai-first-llmisvc.md: added "New to KServe?" info callout pointing new users to the simpler tutorial firstswagger-ui.md: added callout linking to full V1/V2 protocol documentationv1-protocol.md,v2-protocol.md: standardized titles andsidebar_labelfrontmatterReviewer Notes
intro.md → intro.mdxrename is intentional: MDX is required for<Tabs>,<TabItem>, and JSX grid cardsTest plan
npm run buildpasses with no broken link warningsintro.mdxresolve (no 404s)🤖 Generated with Claude Code