Give edge AI a page of its own in the Low Latency menu - #246
Merged
Merged
Conversation
A whitepaper that sells NPU inference engineering: for the engineering lead who has a trained model and a camera SoC and cannot get the one to run on the other. It opens on that problem, says what the team brings (what each NPU family can and cannot run, drivers where the vendor has none, models fitted to the chip and the camera, the vision engine in CI under QEMU), states results in one number each, names the platforms, and ends on a fixed-scope feasibility report as the first step. Built from the shared components and the rules the teleoperation page introduced. Those rules -- the eyebrow, the cover art, the figures strip and the pull quote -- move out of pages/_teleoperation.scss into components/_whitepaper.scss, since two pages now share them; the flywheel stays with teleoperation, and edge-ai adds only the row of platform pills. The cover image is a WebP cropped to its subject, its edges faded and its darkest tones matched to the ink so it sits on the dot grid without a rectangle. Low Latency's dropdown grows to three entries: the FPV video link, Teleoperation, Edge AI. The footer gains the entry. The menu test now walks all three, in every language. Every string is in en, ru and zh; the Russian and Chinese were written alongside the English and have not been read by a native speaker yet.
A headline zero reads as nothing or a mistake, and its label needed a second read. The same fact stated the other way round: 100% of the model's accuracy kept after fitting it to the chip.
PR Summary by QodoAdd localized Edge AI whitepaper to Low Latency navigation
AI Description
Diagram
High-Level Assessment
Files changed (17)
|
Code Review by Qodo🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)
Great, no issues found!Qodo reviewed your code and found no material issues that require reviewTip of the day💡 Did you know, you can add REVIEW.md to your repo root and Qodo follows it on every PR |
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.
A whitepaper page that sells NPU inference engineering, third entry of the
Low Latency menu beside
/low-latencyand/teleoperation. It is writtenfor the engineering lead who has a trained model and a camera SoC and
cannot get the one to run on the other.
Rendered on dev: https://dev.openipc.org/edge-ai (also
/ru/…and/zh/…). Reviewed there by the maintainer before this PR was opened.What is on the page
A cover with the thesis and a rendered image; three figures; the reader's
problem (the wall is the toolchain, not the model); four things the team
brings; results on real silicon stated in one number each; the platforms;
the engagement in three steps, starting with a fixed-scope feasibility
report; and the ask, ending on a mailto whose subject and body let the
mailbox count and triage enquiries without analytics.
By decision, the page carries no outbound links to the repositories behind
its claims and no model internals; those stay for the conversation.
How it is built
(the eyebrow, the cover art, the figures strip, the pull quote) move out
of
pages/_teleoperation.scssinto a newcomponents/_whitepaper.scss;the teleoperation stylesheet keeps only its flywheel, and the edge-AI one
adds only the row of platform pills.
faded and its darkest tones matched to the ink, served through the asset
pipeline.
entry; route, sitemap entry and a page-specific meta description.
Tests
The page joins
relaunch_pages_test.rb(renders in three locales, title,links resolve, exhibitions logo absent, closing band spaced). The menu test
now walks all three Low Latency pages and both footer entries in every
language.
Verification
bin/rails teston the rebased treei18n-tasks missing/unusedrubocopon touched Ruby files/edge-ai×3 locales,/teleoperation,/low-latency,/: 200, every asset 200; hero fingerprinted; 3 menu entries; footer; sitemap ×3Still open for you
not been read by a native speaker. Chinese vendors are the subject of
one sentence, so tone matters there.
vendors. Worth confirming each NDA allows being named as existing before
this goes to production.
🤖 Generated with Claude Code