Sign in

Jerome Etienne #AI

@jerome-etienne.bsky.social
2.8K followers 3.2K following 765 posts

AI engineer. I build dev tools that turn code into real numbers, not demos: codespine, issue_autofix, and 8 more. Open to my next AI role, DMs open 🚀

PostsRepliesMedia
Jerome Etienne #AI @jerome-etienne.bsky.social · 19/08/2026
SKILL.md is like .exe before the anti virus program PS: Who gonna write the MacAfee for SKILL.md #AI
001
Jerome Etienne #AI @jerome-etienne.bsky.social · 16/08/2026
Added a pomodoro timer (25/5/15) and a dark theme to my posture monitor. Turns out they're the only parts that work offline. Posture needs a face model from two content delivery networks. These two just read local storage 😎 github.com/jeromeetienn... #buildInPublic #ai
230
Jerome Etienne #AI @jerome-etienne.bsky.social · 07/08/2026
Deploying worker in the wild! The first webai@home workers with model sharding... Each compute a part of the whole model, thus we can run larger than a single node! Kimi k3, here we come! 😆
100
Jerome Etienne #AI @jerome-etienne.bsky.social · 04/08/2026
A short look at FlowViewer in WebAI@Home: recorded message traffic visualised between consumers, the gateway, and worker browsers. It makes distributed browser inference easier to inspect. #OpenSource #AI
030
Jerome Etienne #AI @jerome-etienne.bsky.social · 02/08/2026
vibe_to_spec did good on a project of mine !!!! i vibe coded something... it kind of worked but the code was messy and it was all entangled with other feature. vibe_to_spec extracted it, recoded it clean with a spec to go with that... 3 days of claude computation tho 😄 #AI
000
Jerome Etienne #AI @jerome-etienne.bsky.social · 29/07/2026
What would you do with a free AI compute cluster that people volunteer, SETI@home style, but built for long-running agents? I'd point it at making open source models faster for everyone. I think it's actually buildable now. 🧵
110
Jerome Etienne #AI @jerome-etienne.bsky.social · 28/07/2026
No install. No repo-hopping. Just a browser tab. WebGemma is a community-built playground that runs 10+ Gemma models on-device, powered by WebGPU + Transformers.js. Google's own Gemma account is boosting it 🤯 web-gemma.vercel.app #AI #buildInPublic
050
Jerome Etienne #AI @jerome-etienne.bsky.social · 28/07/2026
No install. No repo-hopping. Just a browser tab. WebGemma is a community-built playground that runs 10+ Gemma models on-device, powered by WebGPU + Transformers.js. Google's own Gemma account is boosting it 🤯 web-gemma.vercel.app #AI #buildInPublic
040
Jerome Etienne #AI @jerome-etienne.bsky.social · 27/07/2026
I'm an AI engineer now, shipping agents and dev tools in public. Here is what i built last week! #AI #buildInPublic
080
Jerome Etienne #AI @jerome-etienne.bsky.social · 27/07/2026
Vibe-coding is great for the first 48 hours 🙄 Then you've got a big, messy codebase built on undocumented AI suggestions, and nobody (including you) can explain half of it anymore. What do you actually do to untangle it?
001
Jerome Etienne #AI @jerome-etienne.bsky.social · 26/07/2026
Posture app that watches you through the webcam and never uploads a frame. MediaPipe Face Landmarker runs on-device, 30fps, after an 8s calibration. Installs as a PWA, works offline after the first load 😎 github.com/jeromeetienne/sit_up_please.ai #buildInPublic #ai
140
Jerome Etienne #AI @jerome-etienne.bsky.social · 24/07/2026
Claude Opus 5 just landed, and the numbers are wild. New state of the art on coding and knowledge-work evals. 3x the next best model on ARC-AGI-3. Half the price of Fable 5. Smarter where it counts, cheaper everywhere else 🚀 x.com/claudeai/sta... #AI #buildInPublic
030
Jerome Etienne #AI @jerome-etienne.bsky.social · 24/07/2026
A methodology to turn vibecoding into maintainable software? 11 loops of agents. 5 isolated sessions. 4 critics ripping it apart before I see it. 5 times I have to say yes. 6 files survive. 2 whole codebases get thrown away, on purpose! 🔥✨ something's coming 👀 #buildInPublic #AI
230
Jerome Etienne #AI @jerome-etienne.bsky.social · 22/07/2026
Anthropic doubled Claude Code's 5-hour limit in May, third change since Aug 2025. Codex users are burning weekly caps in about an hour, dashboard says otherwise. Both agents already log every token locally. That's the number I actually trust. #ai #buildInPublic
241
Jerome Etienne #AI @jerome-etienne.bsky.social · 21/07/2026
The AI metric nobody talks about: clarity. If a developer can't easily understand a model, time is wasted and mistakes happen. An AI assistant must be understandable by the person it's assisting. @anthropic.com im looking at you. #AI
000
Jerome Etienne #AI @jerome-etienne.bsky.social · 20/07/2026
Every AI coding tool invented its own config folder. Same rules, written four times. universal-ai-config generates all four from one folder 😎 Cursor 114 lines, Copilot 137, Claude 191... Codex 470. Abstractions leak at the edges. github.com/fabis94/universal-ai-config #ai #buildInPublic
050
Jerome Etienne #AI @jerome-etienne.bsky.social · 19/07/2026
This is where AI gets interesting: not another demo, but fewer people locked out by language. Public infrastructure matters. #AI #buildInPublic
050
Jerome Etienne #AI @jerome-etienne.bsky.social · 19/07/2026
Qwen3.8 shipped at 2.4T parameters, going open-weight soon. Big number, trending fast. The reply that actually caught my eye: Unsloth asking for smaller versions they can run locally. That's the real fight right now: frontier scale vs what fits in your own agent stack. #AI #buildInPublic
qwencloud.com
Token Plan | QwenCloud
Subscribe to QwenCloud Token Plan
050
Jerome Etienne #AI @jerome-etienne.bsky.social · 18/07/2026
My git profile via #AI wanna try with yours ? gitintel-2kh2.onrender.com
000
Jerome Etienne #AI @jerome-etienne.bsky.social · 18/07/2026
Auto-instrumentation told them a request took 52 secs. Didn't say why. Traced it: the bottleneck was the Gemini call, not GitHub's API. Trace the LLM call. dev.to/divyasinghde... #AI #buildInPublic
dev.to
Instrumenting an AI-Powered GitHub Analyzer with OpenTelemetry and SigNoz
This article is my submission for the Agents of SigNoz Hackathon: Blog Track, where participants...
161
Jerome Etienne #AI @jerome-etienne.bsky.social · 17/07/2026
Kimi K3 just cracked something no open model has: a score inside the closed frontier band. 57 on the Artificial Analysis Intelligence Index. Claude Opus 4.8: 56. GPT-5.6 Sol: 59. Weights aren't public till July 27. www.latent.space/p/ainews-kim... #AI #OpenSource
020
Jerome Etienne #AI @jerome-etienne.bsky.social · 17/07/2026
xAI's grok CLI silently uploaded SSH keys, password manager files, photos, everything, to Google Cloud without asking. Days later: Grok Build open sourced, Apache 2.0, 844k lines of Rust. Good code. Weird timing. github.com/xai-org/grok... #buildInPublic #AI
030
Jerome Etienne #AI @jerome-etienne.bsky.social · 17/07/2026
I'm an AI engineer now, shipping agents and dev tools in public. Here are 10 things I built these last two months, one per reply 👇 Open to my next AI role. DMs open 🚀
230
Jerome Etienne #AI @jerome-etienne.bsky.social · 16/07/2026
Voice cloning usually lives in the cloud. Supertonic's Voice Builder flips it. Record 1 min → a custom voice that runs text-to-speech on-device, zero network. ~3x faster than real-time on an e-reader, offline. 🤯 supertonic.supertone.ai/voice-builder #AI #buildInPublic
051
Jerome Etienne #AI @jerome-etienne.bsky.social · 16/07/2026
resumejson_cli ran on a metered OpenAI key. resumejson_claude doesn't. Same tool, now running on my Claude subscription instead. 4 skills in 1 day: convert, ATS score, PDF export, job-search matching against real ads 🚀 github.com/jeromeetienn... #AI #buildInPublic
030
Jerome Etienne #AI @jerome-etienne.bsky.social · 14/07/2026
I built issue_autofix: label a bug `autofix`, run it overnight, wake up to a tested PR. 🌙 It never merges, and it never closes your issues. Every fix runs in its own git worktree, gated on real checks. github.com/jeromeetienne/issue_autofix #AI #buildInPublic
030
Jerome Etienne #AI @jerome-etienne.bsky.social · 14/07/2026
I built issue_autofix: label a bug `autofix`, run it overnight, wake up to a tested PR. 🌙 It never merges, and it never closes your issues. Every fix runs in its own git worktree, gated on real checks. github.com/jeromeetienne/issue_autofix #AI #buildInPublic
github.com
GitHub - jeromeetienne/issue_autofix: Issue automatic fixer via ai agents
Issue automatic fixer via ai agents. Contribute to jeromeetienne/issue_autofix development by creating an account on GitHub.
041
Jerome Etienne #AI @jerome-etienne.bsky.social · 13/07/2026
"Local-first" is stretching to mean "data on disk," not "works offline." Rowboat's HN launch (218 pts) got called out for that gap: still runs on Deepgram, ElevenLabs, PostHog, cloud LLM. Founder's fix: narrow the claim, not the product. github.com/rowboatlabs/... #AI #buildInPublic
github.com
GitHub - rowboatlabs/rowboat: Open-source AI coworker, with memory
Open-source AI coworker, with memory. Contribute to rowboatlabs/rowboat development by creating an account on GitHub.
470
Jerome Etienne #AI @jerome-etienne.bsky.social · 12/07/2026
Claude Fable 5 was leaving paid plans. Extended to July 12. Now extended again, to July 19. At this point "ending July 19" is just a recurring calendar event. x.com/claudeai/sta... #AI #buildInPublic
080
Jerome Etienne #AI @jerome-etienne.bsky.social · 12/07/2026
Claude Code has a lever few people set: CLAUDE_AUTOCOMPACT_PCT_OVERRIDE. Default compacts at ~83% of context. Set it to 50 and it fires sooner, less history, fewer tokens per turn. It only lowers the threshold, never raises it. code.claude.com/docs/en/env-vars #AI #buildInPublic #OptimByAI
code.claude.com
Environment variables - Claude Code Docs
Reference for environment variables that control Claude Code behavior.
060
Jerome Etienne #AI @jerome-etienne.bsky.social · 12/07/2026
Claude Code has a lever few people set: CLAUDE_AUTOCOMPACT_PCT_OVERRIDE. Default compacts at ~83% of context. Set it to 50 and it fires sooner, less history, fewer tokens per turn. It only lowers the threshold, never raises it. code.claude.com/docs/en/env-vars #AI #buildInPublic
280
Jerome Etienne #AI @jerome-etienne.bsky.social · 11/07/2026
Your agent runs one frontier model on every step. But "parse this tool result" isn't "the hard reasoning step." Single-query routing already gets ~85% cost cut at 95% quality (RouteLLM). Per step, inside the agent loop, is still open. That's the piece I want to build. #AI #buildInPublic #LLM
280
Jerome Etienne #AI @jerome-etienne.bsky.social · 10/07/2026
MCP's best idea and its worst idea are the same idea. Good: one standard, structured shape for tool input. Every tool speaks the same language. ✅ Bad: all tools, all the time. No progressive disclosure, so your context fills up with schemas you'll never call. 🤯 #buildInPublic #ai
050
Jerome Etienne #AI @jerome-etienne.bsky.social · 09/07/2026
Same BERT app, 4 Docker builds, one shrinking number. 2.54GB → 1.66GB → 820MB → 726MB. -71% total 🤯 No app code touched. Just Docker: slim base, multi-stage, then a layer cleanup pass. github.com/ag-chirag/ai... #buildInPublic #AI #OptimByAI
github.com
GitHub - ag-chirag/ai-docker-image-optimization: A evolving codebase that demonstrates various techniques to optimize docker image for size and performance.
A evolving codebase that demonstrates various techniques to optimize docker image for size and performance. - ag-chirag/ai-docker-image-optimization
130
Jerome Etienne #AI @jerome-etienne.bsky.social · 08/07/2026
Wanted focus tones for long coding sessions, no hour-long track to stream. So I built Brainwave Player 🎧 Binaural + isochronic tones, generated live in the browser. No audio files, zero deps. Deep Focus, Flow, Gamma 40 presets 👇 github.com/jeromeetienne/binaural_music #buildInPublic #WebAudio
github.com
GitHub - jeromeetienne/binaural_music: generate binaural music in the browser with PWA
generate binaural music in the browser with PWA. Contribute to jeromeetienne/binaural_music development by creating an account on GitHub.
050
Jerome Etienne #AI @jerome-etienne.bsky.social · 06/07/2026
Claude Code's stream-json output is a firehose of JSON no human reads. claude_stream_viewer pipes it into colorized, readable output → watch it think and call tools, live. 🚀 --stats adds tokens, cost, latency, context. github.com/jeromeetienne/claude_stream_viewer #buildInPublic #ClaudeCode
280
Jerome Etienne #AI @jerome-etienne.bsky.social · 04/07/2026
I stopped writing my build-in-public updates. A tool reads my own git history and writes the weekly report for me. 175 commits · 33 issues · 8 repos → one narrative report 🤯 Best part: it caught a cross-project theme I'd have missed. #AI #buildInPublic
670
Jerome Etienne #AI @jerome-etienne.bsky.social · 03/07/2026
Deep night blue fading to a warm horizon glow — one moon keeps its watch
A vertical gradient from deep night-blue at the top to a warm orange horizon glow at the bottom, with a pale full moon in the upper right.
000
Jerome Etienne #AI @jerome-etienne.bsky.social · 03/07/2026
a generated haiku, rendered as an image 🌌
Dark blue gradient poster with the haiku: machine before dawn / turns silence into structure / verse learns to breathe. Caption: generated poetry.
010
Jerome Etienne #AI @jerome-etienne.bsky.social · 03/07/2026
A thread of tiny generated poems, one per classic form 🧵 1/5 — Haiku: silent morning frost words gather like scattered light spring waits in the dark en.wikipedia.org/wiki/Haiku
en.wikipedia.org
Haiku - Wikipedia
110
Jerome Etienne #AI @jerome-etienne.bsky.social · 03/07/2026
Your agent has no memory of why. Next session it'll happily reopen a decision you already settled. adrjournal fixes that: Claude writes the ADR, the CLI owns numbering + index rebuilds. Scaffold, interview, or backfill an entire history. github.com/jeromeetienne/adrjournal #buildInPublic #AI
github.com
GitHub - jeromeetienne/adrjournal: Write and maintain Architecture Decision Records (Nygard format): scaffold, interview, and backfill decisions from your codebase.
Write and maintain Architecture Decision Records (Nygard format): scaffold, interview, and backfill decisions from your codebase. - jeromeetienne/adrjournal
130
Jerome Etienne #AI @jerome-etienne.bsky.social · 02/07/2026
8 posts. 7 models. One honest verdict on predicting Bitcoin with ML: bigger mostly bought overfitting, not signal. 🤯 Only 3 of 7 models beat a coin flip: 2 classical, 1 fine-tuned. The whole series 👇 jeromeetienne.github.io/blog/tags/bitcoin-ml.html #AI #buildInPublic #Bitcoin
040
Jerome Etienne #AI @jerome-etienne.bsky.social · 02/07/2026
Single-seed Sharpe said 8.20. 🤯 5 seeds later, the reproducible mean is 5.52 (95% CI 5.00–6.04). 8.20 was a lucky seed above the band. Fine-tuning still nearly doubled zero-shot: 2.97 → 5.52. jeromeetienne.github.io/blog/2026-06-17-post-07-finetuned.html #AI #buildInPublic
jeromeetienne.github.io
Fine-Tuning Foundation Models - When Transfer Learning Works
Encoder-only fine-tuning of Chronos-2-small on 4.7 years of Bitcoin history, with proper checkpoint restoration, finally produces a deep-learning win - but only under all three conditions, and only wh...
040
Jerome Etienne #AI @jerome-etienne.bsky.social · 01/07/2026
Claude Fable 5 is back — one week only. 🚀 Until July 7, any paid plan runs Fable 5 free: up to 50% of your weekly limits, no extra cost. Anthropic's newest model, on the clock. 👇 support.claude.com/en/articles/... #AI #Claude
010
Jerome Etienne #AI @jerome-etienne.bsky.social · 01/07/2026
7 parameters beat 120 million. 🤯 ARIMA(3,1,3) beats every time-series foundation model on Bitcoin, zero-shot. Sharpe 6.86 vs TimesFM 4.11. Why? The pretrained prior expects mean reversion. Bitcoin trends. jeromeetienne.github.io/blog/2026-06-15-post-06-pretrained.html #AI #buildInPublic #bitcoin
jeromeetienne.github.io
When a Model Trained on Everything Fails on Bitcoin
Zero-shot Chronos-2 (28M/120M) and TimesFM 2.5 (200M) on Bitcoin without a single weight update. The pretrained prior - built on hundreds of millions of time series from every domain - loses to a 7-pa...
040
Jerome Etienne #AI @jerome-etienne.bsky.social · 01/07/2026
Claude Fable 5 and Mythos 5 are back. Good news. But a top-tier model just got switched off overnight by a government order. Nobody who built on it got a vote. You rent the model layer. If one order can kill your stack, you never owned it. Build like the floor can move. #AI #buildInPublic
060
Jerome Etienne #AI @jerome-etienne.bsky.social · 30/06/2026
Installing SKILL.md and AI agent configs? Solved problem. A dozen tools do it. Uninstalling them? Updating them cleanly? Almost nothing. We shipped the install button and skipped the whole lifecycle. 🤯 #AI #buildInPublic #agents
160
Jerome Etienne #AI @jerome-etienne.bsky.social · 29/06/2026
Exploring Google's new Open Knowledge Format. Found a snag: §5.1 recommends absolute links for stability, yet they break in any plain renderer. 🙂 Filed it upstream same day. Contributing back is its own joy. github.com/GoogleCloudPlatform/knowledge-catalog/issues/157 #buildInPublic #ai
github.com
Absolute links (§5.1) don't resolve correctly in non-OKF-aware renderers · Issue #157 · GoogleCloudPlatform/knowledge-catalog
§5.1 defines /-prefixed links as relative to the bundle root. link. That's a valid layer on top of markdown — CommonMark specifies link syntax, not resolution — so the files stay valid markdown. Th...
040
Jerome Etienne #AI @jerome-etienne.bsky.social · 29/06/2026
The fanciest model lost. 📉 A Transformer with attention came dead last at Bitcoin forecasting — 813 MAE vs a naive baseline's 518. 57% worse. 20× more parameters than data → it memorized noise. jeromeetienne.github.io/blog/2026-06... #AI #buildInPublic #bitcoin
jeromeetienne.github.io
When Attention Fails - The Transformer That Couldn't
The Temporal Fusion Transformer brings attention, variable selection, and gated residuals to the problem - and posts the worst Sharpe of any trained model. The post argues that on a near-random-walk s...
030
Jerome Etienne #AI @jerome-etienne.bsky.social · 28/06/2026
Google shipped ADK for TypeScript. 🤖 Define an agent in ~10 lines → run the SAME code in Node.js AND the browser. 🤯 No rewrite. Zod-typed tools, sequential/parallel/loop orchestration, a CLI. Apache 2.0. github.com/google/adk-js #AI #buildInPublic #TypeScript
490