Sign in

Jeremy Morgan

@jeremymorgan.com
399 followers 442 following 507 posts

Developer, Tech Blogger, and Volunteer Firefighter. World's okayest developer "You don't need permission to be awesome" - @reverentgeek.com I came here to break builds and chew bubblegum. And I'm all out of bubble gum.

PostsRepliesMedia
Jeremy Morgan @jeremymorgan.com · 6h
Six Degrees of Ayrton Senna: Learn Neo4j by Connecting 75 Years of Formula 1 dzone.com/articles/neo...
dzone.com
Learn Neo4j by Connecting 75 Years of Formula 1
Learn how to build an F1 teammate graph in Neo4j, derive relationships from race data, and connect Max Verstappen to Juan Manuel Fangio with Cypher.
000
Jeremy Morgan @jeremymorgan.com · 29/09/2026
Raspberry Pi Smart Display Module available now at $30 www.raspberrypi.com/news/raspber...
raspberrypi.com
Raspberry Pi Smart Display Module available now at $30 - Raspberry Pi
The Raspberry Pi Smart Display Module is available to buy now from our global network of Approved Resellers.
000
Jeremy Morgan @jeremymorgan.com · 26/09/2026
tokentab reads the session logs Claude Code, Codex and Gemini CLI already leave on disk and totals cost by model, project and day. Local only. No account, no API key. Cursor support is a stub, and the README says so. github.com/crwdla/token...
github.com
https://github.com/crwdla/tokentab
110
Jeremy Morgan @jeremymorgan.com · 26/09/2026
Capsule packs an HTML app and its data into one SQLite file with a .capsule extension. Entries carry UUIDs and timestamps so diverged copies can merge. Sandboxed, no network by default. Rust and Tauri. The format spec is promised at 1.0, so treat it as early. withcapsule.app
withcapsule.app
Capsule — Documents that run like apps
One file, entire app. Share like a document, open like an app.
000
Jeremy Morgan @jeremymorgan.com · 26/09/2026
FlexViz keeps charts interactive at 100M+ rows by answering every zoom and brush with lazy Polars aggregations and Rust kernels. Nothing raw ships to the browser. Apache 2.0, on PyPI, with a live demo and public benchmarks. Early and small. Worth a look. github.com/flex-analyti...
github.com
GitHub - flex-analytics/flexviz: Interactive data exploration that is extremely fast and agent-native
Interactive data exploration that is extremely fast and agent-native - flex-analytics/flexviz
010
Jeremy Morgan @jeremymorgan.com · 26/09/2026
Redis City: type a command and follow it through a 3D model of one Redis instance. Parsing, key lookup, the allocator, memory pages, RDB and AOF. Components link to the source. A visual explainer that teaches the internals instead of decorating them. poltora.dev/redis
poltora.dev
Redis City — inside one Redis
An interactive 3D Redis visualization, from command parsing and key lookup to the allocator, memory pages, RDB, and AOF.
000
Jeremy Morgan @jeremymorgan.com · 26/09/2026
Z.ai stood up GLM-5.3-Flash serving on 100k+ Chinese accelerators with an infra agent doing much of the work. Throughput tripled in under two weeks. The reusable part is dense feedback: correctness tests, traces and microbenchmarks, not one aggregate number. z.ai/blog/glm-bui...
z.ai
Z.ai - Advanced AI Chatbot & Agent powered by GLM-5.3-Flash
Meet Z.ai, the AI assistant powered by GLM-5.3-Flash. Build websites, write code, handle long-horizon tasks, and get instant answers. Fast, smart, and reliable.
000
Jeremy Morgan @jeremymorgan.com · 26/09/2026
Mark Litwintschik plans a fuel-optimal A320 route with Airbus's scikit-decide and the OpenAP performance model, then exports to Parquet and maps it over wind data in QGIS. Every command is on the page. The HN thread supplies the ETOPS rules the solver ignores. tech.marksblogg.com/scikit-decid...
tech.marksblogg.com
Saving Jet Fuel
Benchmarks & Tips for Big Data, Hadoop, AWS, Google Cloud, PostgreSQL, Spark, Python & More...
000
Jeremy Morgan @jeremymorgan.com · 26/09/2026
pgrust 0.3: a Rust reimplementation of Postgres that passes all 46,066 upstream regression tests, now tracking 18.6. This release is reliability work. One docker run to try it, or use the browser demo. The authors say it is not production ready. Believe them. github.com/malisper/pgr...
github.com
Release v0.3: Try us on real world workloads in non-critical environments · malisper/pgrust
Releasing pgrust v0.3, our Postgres rewrite from C to Rust. This release is mostly about making pgrust more reliable. Try it on real world workloads in non-critical environments. pgrust is not prod...
000
Jeremy Morgan @jeremymorgan.com · 25/09/2026
Polars 2.0 RC2 is out. Breaking changes first: a new Map dtype, cut/qcut deprecated, Parquet ENUM now reads as String. Then the optimizer work: dynamic predicates for hash joins, predicates derived from join conditions. Run your pipelines against it now. github.com/pola-rs/pola...
github.com
Release Python Polars 2.0.0-rc.2 · pola-rs/polars
💥 Breaking changes Read Parquet ENUM type as pl.String (#29331) More map operations (#29296) Deprecate cut/qcut (#29329) Remove legacy POLARS_STREAMING_CHUNK_SIZE (#29046) Introduce the Map dtype ...
000
Jeremy Morgan @jeremymorgan.com · 25/09/2026
DuckDB-Wasm can keep a real database file in the browser's Origin Private File System. Open an opfs:// path and it survives reloads and restarts. No more Parquet-in-IndexedDB workaround. The post is careful about when data actually reaches disk. duckdb.org/2026/09/18/o...
duckdb.org
Persistent Databases in the Browser with DuckDB-Wasm and OPFS
DuckDB-Wasm can open a persistent database file in the browser's Origin Private File System (OPFS). This post shows how, and when data reaches disk.
120
Jeremy Morgan @jeremymorgan.com · 25/09/2026
Agent memory as a deductive database instead of a vector store. Lemmalog is a Rust Datalog engine with bi-temporal facts, provenance on every derivation, why() proof trees, and an MCP server. The README carries an honest status table of what actually shipped. github.com/JordyZomer/l...
github.com
GitHub - JordyZomer/lemmalog: A Datalog engine for LLM agent memory: stratified rules, provenance-tracked facts, incremental derivation, and an MCP server that lets your harness use it as a shared brain.
A Datalog engine for LLM agent memory: stratified rules, provenance-tracked facts, incremental derivation, and an MCP server that lets your harness use it as a shared brain. - JordyZomer/lemmalog
370
Jeremy Morgan @jeremymorgan.com · 25/09/2026
The week's loudest launch: a model that returns typed decisions with probabilities instead of text, in 70 to 500ms. Read the "Nuance" notes. The reference answers are an average of two frontier LLMs and the workflows were written in-house. Credit to them for saying so. typesafe.ai/blog/introdu...
typesafe.ai
Introducing System One Models & Jev - TypeSafe AI Blog
TypeSafe AI is an AI lab building machine-native intelligence infrastructure for automation, designed to make decisions within software. Try our first System One Model, Jev, in early access.
000
Jeremy Morgan @jeremymorgan.com · 25/09/2026
An eBPF security agent was spending its kernel budget working out which policy applied to each file open, not enforcing it. Caching the answer per inode cut CPU cost about 90%. The real lesson is the cache key and invalidation under Linux filesystem semantics. nathannaveen.dev/posts/droppi...
nathannaveen.dev
Dropping eBPF CPU Cost by About 90% With Memoization (Not AI Gen)
My brother and I spent a lot of time designing our eBPF security agent to be really fast from the ground up, but recently we discovered we could make it much faster using memoization! A couple of…
000
Jeremy Morgan @jeremymorgan.com · 25/09/2026
Postgres 19 ships planner hints that live outside SQL. pg_plan_advice reads a plan back as a string. pg_stash_advice pins it by query id with no app change. Dimitri Fontaine runs the full loop on beta 3, including the "matched, failed" feedback most people will miss tapoueh.org/blog/2026/09...
tapoueh.org
Plan Advice in PostgreSQL 19
There is a conversation that happens in every PostgreSQL shop eventually. A query that has been fine for a year gets slow overnight. Nothing was deployed. The …
110
Jeremy Morgan @jeremymorgan.com · 25/09/2026
Anthropic's CI job volume grew 25x in six months as agents pushed small PRs around the clock. Their answer: deterministic test selection built on a per-test history, patched three times before it held. The planning number to steal: assume 25x load in two quarters. claude.com/blog/agentic...
claude.com
Agentic coding is straining CI. Here’s how we scaled test impact analysis at Anthropic | Claude by Anthropic
Our CI job volume increased 25x over 6 months. We patched our test selection service three times before finding a sustainable solution.
000
Jeremy Morgan @jeremymorgan.com · 24/09/2026
Berkeley and Arena ran 7 models through Claude Code, Codex CLI and Pi. Success rates moved a couple of points. Cost moved up to 5x. Same model, same tasks. If you pay per token, the harness is a line item. Read the sample-size limits before you switch. harnesstax.github.io
harnesstax.github.io
HarnessTax: How Much Does the Harness Matter for Coding Agents?
What does a coding-agent harness actually add, and at what cost? It turns out your Claude models may not need Claude Code… We evaluate 21 model–harness pairs spanning seven models and three…
293
Jeremy Morgan @jeremymorgan.com · 24/09/2026
Cloudflare reclaimed 100TB of RAM from a consistent hash ring. Two fixes: repack a Rust struct to kill alignment padding (25%), then cut hashes per server by 90%. The last 90,000 hashes were buying 0.7% better balance. Go look at your own rings. blog.cloudflare.com/saving-100-t...
blog.cloudflare.com
Saving another 100TB of RAM with math (and Rust)
Cloudflare's global network is immense but not limitless. As we look for small ways to trim our resource usage, we sometimes get lucky and we can cut significantly more. Here’s how we reduced one of…
000
Jeremy Morgan @jeremymorgan.com · 24/09/2026
What if AI didn’t need to generate answers at all? Michael Hunger tested Jev by letting it navigate a Neo4j graph one decision at a time. It picked from existing relationships only. The result offers an interesting alternative to the usual LLM approach. neo4j.com/blog/genai/n...
neo4j.com
Navigating a Neo4j Knowledge Graph with Jev - Neo4j Graph Intelligence Platform
Testing Jev (TypeSafe System One) with Navigation Decisons I built neo4jev as an experiment to navigate a graph, here I explain a bit the why and how. Streamlit app in neo4jev that allows database…
000
Jeremy Morgan @jeremymorgan.com · 22/09/2026
TypeSafe's Jev Trades Text Generation for Instant, Calibrated Decisions arcturus-labs.com/blog/2026/09...
arcturus-labs.com
TypeSafe's Jev Trades Text Generation for Instant, Calibrated Decisions
TypeSafe's Jev skips text generation entirely, answering structured questions with calibrated probabilities in milliseconds. Here's how the API works, what might be under the hood, and why it could…
000
Jeremy Morgan @jeremymorgan.com · 21/09/2026
A 4B Qwen, SFT plus RL, steering Postgres through pg_hint_plan: 1.81x geomean on the Join Order Benchmark for $1,200 all in. The best part is the measurement rig. Raising shared_buffers to 2GB cut phantom rewards about 4x. Code is public. rohanbansal.com/qorl
rohanbansal.com
Training a 4B model to produce 81% faster query plans than Postgres
...or how to make Qwen learn query optimization via agentic reinforcement learning
000
Jeremy Morgan @jeremymorgan.com · 09/09/2026
I am at Devopsdays Portland today... are you? Let's meet up!
010
Jeremy Morgan @jeremymorgan.com · 04/09/2026
Someone crammed a surprisingly capable Raspberry Pi Zero cyberdeck into a mint-tin-sized package. This is exactly the kind of unnecessary-but-awesome mini computer that makes maker projects fun. www.raspberrypi.com/news/curious...
raspberrypi.com
Curiously minty Raspberry Pi Zero cyberdeck - Raspberry Pi
Marrying the Raspberry Pi community's enduring love for cyberdecks and building tiny projects inside Altoids tins.
010
Jeremy Morgan @jeremymorgan.com · 04/09/2026
Concrete git worktree workflow for running several branches, or several coding agents, at once without stashing. Short, all commands, and it removes real friction now that parallel agent sessions are normal. barrd.dev/article/para...
barrd.dev
Parallel development without the headaches using Git worktree : barrd.dev
Git worktree, run multiple branches in parallel directories to juggle features, hotfixes and testing without stashing or constant checkouts.
030
Jeremy Morgan @jeremymorgan.com · 04/09/2026
This free interactive ESP32 pinout reference doesn't just show GPIO numbers. It tells you the real constraints and traps attached to each pin, using Espressif's hardware data and community validation. developer.espressif.com/blog/2026/08...
developer.espressif.com
esp32pin.com: a pinout reference that knows which pins will bite you
esp32pin.com is a free interactive pinout reference for the ESP32 family. Every pin carries its real-world constraints, the data is generated from Espressif’s official KiCad libraries and…
010
Jeremy Morgan @jeremymorgan.com · 03/09/2026
SPOKE is an RP2040-powered touch controller designed for music and other interactive applications. It's the kind of project that makes capacitive touch feel much more interesting than a simple on/off button. www.raspberrypi.com/news/spoke-a...
raspberrypi.com
SPOKE: An RP2040-powered touch controller - Raspberry Pi
SPOKE is a well-designed RP2040-powered touch controller that can also act as a MIDI input or HID.
020
Jeremy Morgan @jeremymorgan.com · 03/09/2026
Argument that inference engines are an underexamined attack surface, and that a model able to influence its own decoding path may be able to reach the host. Speculative in places, but worth having the threat model before you self-host serving infra. boydkane.com/essays/llms-...
boydkane.com
LLMs could control their host machines by exploiting inference engines
| Read on LessWrong | Comment on HN | Large language models often take actions running on one computer (via an agentic harness such as Claude Code or Codex), however the LLMs’ responses to prompts…
000
Jeremy Morgan @jeremymorgan.com · 03/09/2026
Geolocating an unnamed island from a single photo using horizon geometry and a brute-force CUDA search across coastline data. Overkill in the best way, and a good demonstration of when to stop being clever and just parallelize the search. yassa9.github.io/osint/gralhi...
yassa9.github.io
gralhix #004
just a blog
000
Jeremy Morgan @jeremymorgan.com · 03/09/2026
Want a full-size train departure board at home? This Raspberry Pi build drives multiple HUB75 LED matrices and pulls live rail schedule data to recreate the real thing. anotherpartialsuccess.com
anotherpartialsuccess.com
Raspberry Pi UK Train Departure Board Build - Another Partial Success
How to build your own UK Train Departure Board using a Raspberry Pi, some HUB75 LED Matrices and the free software from anotherpartialsuccess
000
Jeremy Morgan @jeremymorgan.com · 02/09/2026
Instead of putting a level sensor inside the tank, this maker clamps a current sensor around the burner wiring. An ESP32 tracks burner runtime and estimates oil use, gallons remaining, burn rate, and days left. www.reddit.com/r/homeautoma...
reddit.com
https://www.reddit.com/r/homeautomation/comments/1vukkwi/i_built_a_heating_oil_monitor_that_warns_me_about/
000
Jeremy Morgan @jeremymorgan.com · 02/09/2026
This guy points a projector at his ceiling and uses a $30 radio dongle to draw every airplane flying over his house in real time. Sun, moon, stars, and the ISS are in there too. MIT licensed: github.com/cpaczek/skyl...
github.com
GitHub - cpaczek/skylight: Project the aircraft passing overhead onto your ceiling in real time, from an RTL-SDR — with a live sky layer (sun, moon, stars, ISS) and where each plane is headed.
Project the aircraft passing overhead onto your ceiling in real time, from an RTL-SDR — with a live sky layer (sun, moon, stars, ISS) and where each plane is headed. - cpaczek/skylight
000
Jeremy Morgan @jeremymorgan.com · 02/09/2026
Durable streams over plain HTTP, backed by object storage. No brokers to run, no partitions to rebalance. The interesting question it raises is how many queues in your stack are actually just append-only logs you are overpaying to operate. picomq.com
picomq.com
PicoMQ
PicoMQ is durable, real-time streams over HTTP, built on S3-compatible object storage.
010
Jeremy Morgan @jeremymorgan.com · 02/09/2026
The argument that per-token price now shapes architecture more than capability does. Once the best model costs enough, you route work by tier instead of upgrading everything. A useful frame if you are designing multi-model systems this quarter. www.dbreunig.com/2026/08/23/f...
dbreunig.com
Fable & The End of the Free Lunch
There’s some talk today about how agentic coders are balking at Anthropic’s pricing and adopting alternatives. I was reminded of a thought I had in the weeks following Fable’s release: the free lunch…
110
Jeremy Morgan @jeremymorgan.com · 02/09/2026
Arduino's Zephyr-based core has officially left beta. It now supports boards including UNO Q, Portenta, GIGA, Nano Matter, Nano RP2040 Connect, Nano 33 BLE, and several Nicla boards using familiar Arduino APIs on top of Zephyr. blog.arduino.cc/2026/07/29/a...
blog.arduino.cc
Arduino Core on Zephyr 0.90.0 is officially stable and leaving beta | Arduino Blog
We are thrilled to announce the release of version 0.90.0 of the ArduinoCore-zephyr. This is a major milestone: we are officially exiting the beta phase and marking the core as stable. Achieving this…
000
Jeremy Morgan @jeremymorgan.com · 01/09/2026
This ESPHome build combines an ESP32 with CO₂, particulate, VOC, NOx, temperature, humidity, and gas sensors, plus a color display, Bluetooth proxying, calibration controls, and Home Assistant integration. www.reddit.com/r/Esphome/co...
reddit.com
https://www.reddit.com/r/Esphome/comments/1vifez0/new_esphomebased_environment_sensor/
010
Jeremy Morgan @jeremymorgan.com · 01/09/2026
A Rust language server claiming 100x less RAM than rust-analyzer. Early, but the design notes on what it deliberately skips are the useful part, especially if you have watched an LSP eat 8GB on a large workspace. rust-glancer.github.io/blog/hello-w...
rust-glancer.github.io
Hello, world! · Rust Glancer
Rust LSP that doesn't eat memory for breakfast
000
Jeremy Morgan @jeremymorgan.com · 01/09/2026
Raspberry Pi published a practical walkthrough for running LiteRT models locally, including language models and image classification. There's even experimental WebGPU support to play with. www.raspberrypi.com/news/get-sta...
raspberrypi.com
Get started with LiteRT on Raspberry Pi - Raspberry Pi
Install and run high-performance on-device AI using the LiteRT CLI tool on Raspberry Pi.
000
Jeremy Morgan @jeremymorgan.com · 31/08/2026
This ESP32 project uses an ultrasonic sensor to monitor a rooftop tank and automatically runs the pump when water gets low. The builder even designed it so the installation could be reversed easily in a rental. hackaday.com/2026/08/02/a...
hackaday.com
An Automated Water Tank Controller
[Jumpor] lives in a part of India where the water infrastructure is a little frustrating to use. Water gets delivered to underground tanks outside of homes, and must then be regularly pumped into r…
000
Jeremy Morgan @jeremymorgan.com · 31/08/2026
Dan Luu on why software slowness is a choice now. Latency measurements across common tools, the usual excuses, and where the costs actually come from. The piece to forward when someone claims the framework is why it takes 400ms. danluu.com/perf-opt/
danluu.com
There's no reason for software to be slow anymore
The other day, I saw a viral tweet saying that people talking about how LLMs are causing slow, bloated, code are going to eat crow once they re-write everything in super-optimized assembly. We're not…
000
Jeremy Morgan @jeremymorgan.com · 31/08/2026
EVE Online is finally leaving Stackless Python 2.7 after 16 years. 2.4 million lines through futurize, then manual review of roughly 20,000 places where Python 2 and 3 semantics diverge. The most honest large migration plan you will read this year. www.eveonline.com/news/view/th...
eveonline.com
The Move to Python 3 Begins! | EVE Online
How do you migrate 2.4 million lines of code?
010
Jeremy Morgan @jeremymorgan.com · 31/08/2026
Simon Willison had an agent build a shot-scraper-style JSON API on Bun 1.4's new WebView, then measured it under cgroups. Full Chrome against complex pages needs 192MB to 256MB. Rare to get an actual memory number attached to a browser automation idea. github.com/simonw/resea...
github.com
https://github.com/simonw/research/tree/main/bun-webview-json-api
000
Jeremy Morgan @jeremymorgan.com · 31/08/2026
Bun 1.4 lands the Rust rewrite plus Bun.WebView, Bun.cron, Bun.markdown, parallel test running, and bun audit fix. 5x lower idle CPU and 50% faster startup on Linux. The WebView addition quietly puts CDP browser automation into the runtime itself. bun.com/blog/bun-v1.4
bun.com
Bun 1.4
Bun 1.4 rewrites Bun in Rust, ships built-in headless browser automation (Bun.WebView), Bun.Image, Bun.markdown, JSON5, JSONL, Terminal and cron APIs, Node.js 26.3.0 compatibility with 1,517 newly…
110
Jeremy Morgan @jeremymorgan.com · 31/08/2026
The companion argument. Most agent failures people blame on the model are harness failures: bad tool descriptions, no verification step, context that rotted three turns ago. Practical framing for anyone actually shipping agents rather than demoing them. code.joejag.com/2026/your-ag...
code.joejag.com
Your Agent Is Not the Model
Software development blog of Joe Wright
000
Jeremy Morgan @jeremymorgan.com · 31/08/2026
A clean definition of the layer everyone keeps conflating with the model: context assembly, tool routing, retries, verification loops. Useful vocabulary when you are debugging agent behavior and cannot tell whether the model or your scaffolding is at fault. earendil.com/posts/what-i...
earendil.com
What is a Harness? | EARENDIL
An agent harness is a piece of software that provides an environment for an AI model to operate within.
101
Jeremy Morgan @jeremymorgan.com · 30/08/2026
Theremin That's Basically Just an ESP32 This clever build uses the ESP32's capacitive-touch hardware to sense hand distance from two antennas, controlling pitch and volume without needing a complicated traditional Theremin circuit. hackaday.com/2026/08/02/e...
hackaday.com
Easy Theremin Uses ESP32
The Theremin is that classic electronic musical instrument that makes those weeee-ohhhhh noises which were so popular in mid-century science fiction movies. Mid-last-century, that is, because this …
000
Jeremy Morgan @jeremymorgan.com · 30/08/2026
This Arduino Nano music player hides RFID tags inside small toys. Put a toy on the wooden player and it automatically starts the matching playlist from an SD card. hackaday.com/2026/08/03/c...
hackaday.com
Child-Friendly Music Player Uses RFID
[David] has a young child who is clever enough to use a computer to play music, but he doesn’t quite want to hand over the mouse just yet. Thus, he set about building an electronic music play…
020
Jeremy Morgan @jeremymorgan.com · 29/08/2026
ESP RainMaker Neo is interesting because Espressif didn't just open-source device firmware. The new implementation exposes the device-to-cloud-to-app stack, including its AWS-based backend, under Apache 2.0. developer.espressif.com/blog/2026/08...
developer.espressif.com
Open Sourcing ESP RainMaker Neo
ESP RainMaker Neo is a new implementation of ESP RainMaker in which the entire device-to-cloud-to-app stack, including the cloud backend, is open source under the Apache License 2.0. This article…
010
Jeremy Morgan @jeremymorgan.com · 29/08/2026
This project replaces the Game Boy Advance audio path with a custom PCB that adds better filtering, amplification, an audio codec, and ESP32-powered Bluetooth — while keeping the case looking stock. github.com/cajunpanda/g...
github.com
https://github.com/cajunpanda/gameboy-hifi-audio/tree/main
000
Jeremy Morgan @jeremymorgan.com · 29/08/2026
PolyShot is a homemade digital camera built around an ESP32-S3-based UNIHIKER K10, a custom button PCB, battery, screen, camera, and a nicely designed enclosure. hackaday.io/project/2063...
hackaday.io
PolyShot Point and Shoot Camera
Greetings, everyone, and welcome back! Here's something fun: meet PolyShot, a DIY point-and-shoot camera that I built from scratch. Lately, point-and-shoot cameras have been making a comeback. My…
020
Jeremy Morgan @jeremymorgan.com · 29/08/2026
Arduino Maker Workshop is an open-source VS Code extension built around Arduino CLI. It handles board selection, libraries, compiling, uploading, profiles, and IntelliSense without forcing you into a completely separate embedded IDE. github.com/thelastoutpo...
github.com
GitHub - thelastoutpostworkshop/arduino-maker-workshop: The ultimate tool for makers to bring Arduino projects to life in Visual Studio Code
The ultimate tool for makers to bring Arduino projects to life in Visual Studio Code - thelastoutpostworkshop/arduino-maker-workshop
000