Jim Jonah

Principal Software Engineer @ Veracode. Design and build cloud native applications. If not programming: boats, motorcycle, photography, or travel.

[Neo Kim @systemdesignone]

I created a playbook to help you learn AI engineering.
It'll give you:

• Core concepts behind AI systems.
• Condensed notes to understand full-stack AI engineering.
• Must know techniques to build, deploy & scale AI apps.

(24 HOURS ONLY!!!)

1 Follow @systemdesignone [MUST]
2 Like & Retweet to get DM
3 Reply "Playbook"

Then I'll DM you the details.

[Jim Jonah @JimJonah]

@systemdesignone Playbook

[Jim Jonah @JimJonah]

@systemdesignone Playbook

[Neo Kim @systemdesignone]

I just created a playbook to help you learn AI engineering.

It gives you:

• Core concepts behind modern AI systems.
• Condensed notes to understand AI engineering stack.
• Must know techniques to build, deploy, debug & scale AI apps.

(24 hours only!)

To get it for FREE:
1. Like, Retweet & Follow @systemdesignone
2. Reply "Playbook"

Then I'll DM you the details.

[Jim Jonah @JimJonah]

@openmarmot @GergelyOrosz Standup is supposed to be less than 5 minutes per person on the team.

When when the tram was too large, we gave two minutes. Someone used a stop watch to keep us on track.

What I did yesterday.
What I’m doing today.
Here’s my roadblocks.
This needs discussion.
Next.

[Jim Jonah @JimJonah]

@danshipper @every I love how this is pulling in lessons from influential business books anti bing to life a very cool system. I will be trying this!

[Alex Kessinger @voidfiles]

This is a great idea, I’ve been doing this in some form for over a decade. My Staff Eng co-host @davidnoelromas reached out this week to ask for more details on how I’ve been using obsidian and AI. This an expanded version of what I told him.

I’ve collected possibly too many markdown files.

> find . -type f | wc -l
52447

That’s my obsidian vault, and I use it with AI everyday without a special database, or a vector store, or a RAG pipeline. It’s merely files on disk.

The problem this actually solves

Think about the context you carry around in your head for your job. The history of decisions on a project. What you discussed with your manager three months ago. The Slack thread where the team landed on an approach. The Google Doc someone shared in a meeting you half-remember. The slowly evolving understanding of how a system works that lives across fifteen people’s heads and nowhere else.

Now think about what happens when you need to produce something from all that context. A design doc. A perf packet. A project handoff. An onboarding guide for a new team member. You spend hours reassembling context from Slack, docs, emails, your own memory, and you still miss things.

The knowledge base turns this into a system instead of a scramble.

The Architecture

A file system with markdown and wikilinks is already a graph database. Files are nodes. Wikilinks are semantic edges. Folders introduce taxonomy. You don’t need a special MCP server or plugin. The file system abstraction is the interface, and LLMs are surprisingly good at navigating it.

I use a structure borrowed from Tiago Forte’s Building a Second Brain, with the PARA taxonomy as a starting point, extended with categories that match how I actually work:

/projects/{name}
/areas/{topics}
/people/{slack_handle}
/daily/{year}/{month}/{day}/
/meetings/{year}/{month}/{day}/

Markdown files are nodes, wikilinks ([[target]]) are edges, the folder taxonomy is the schema and LLMs is the query engine. A graph database with a natural language query interface. No infrastructure required.

How it works day to day

After every meeting, the agent creates a note in daily/{year}/{month}/{day}/, downloads any attached Google Docs, and links everything to the long-running notes I keep for each person I interact with regularly. A note from a 1:1 with my boss JP gets a wikilink to [[/people/jp|jp]] and to whatever projects we discussed.

Over months, each person’s note becomes a timeline of every conversation, decision, and open thread. Each project folder accumulates every relevant artifact. You don’t have to remember where things are. The graph remembers.

For a work project, I can point the agent at a starting doc and say:

> Spider through every tool you have access to and pull down all the related context.

It grabs Slack threads, Google Docs, web resources, all rendered as markdown inside the project folder. From that assembled context, the agent can draft design docs, product vision statements, problem/solution analyses. The output is better than prompting cold because the LLM is working with the real history of the project, not your summary of it.

This is the part Karpathy’s tweet hints at but doesn’t fully spell out: the knowledge base isn’t just for research. It’s a context engineering system. You’re building the exact input your LLM needs to do useful work.

What makes this different from just using an LLM
You might be thinking: I already ask Claude to help me write a design doc. True. But there’s a real difference between prompting “help me write a design doc for a rate limiting service” and prompting an LLM that has access to your project folder with six months of meeting notes, three prior design docs, the Slack thread where the team debated the approach, and your notes on the existing architecture.

The knowledge base is a context engineering system. You’re not building a wiki for the sake of having a wiki. You’re building the input layer that makes every future LLM interaction better. Every meeting note, every linked decision, every filed artifact improves the quality of every query that follows.

Where this is still hard

The piece I haven’t cracked is automated inbox processing. The idea is straightforward: web clippings, meeting notes, Slack saves, and random captures all land in an inbox folder. The agent processes everything new, applies progressive summarization, breaks content into atomic pieces, correlates each piece with the right project, area, or person.

I have a graveyard of experiments here. The LLM is good at summarizing and categorizing. The hard part is defining what “processed” means in a way that’s consistent enough to be useful six months later but flexible enough to handle the variety of stuff that lands in an inbox. Every attempt has been either too rigid (everything gets the same treatment) or too loose (the vault drifts into chaos).

If you’ve solved this, I’d genuinely like to hear about it.
Getting started

You don’t need 52,000 files to get value from this. Start with three things:

1: Create the folder structure. Projects, areas, people, daily. Even empty, the taxonomy gives you and the LLM a schema.

2: After your next meeting, have the agent create a note and link it to the relevant person and project. Do this for a week. Watch the graph start to form.

3: The next time you need to write something, a design doc, a status update, a perf self-review, point the agent at the relevant folders and ask it to draft from what’s there.

The difference is noticeable right away. Not because the LLM is smarter, but because it finally has the context to be useful.

Your work compounds. That’s the thing that feels genuinely new.

[Heinrich @arscontexta]

https://t.co/JnIR1JG33B

[Mario @PawlowskiMario]

Here is my opinion on Minneapolis killing by ICE officer, and facts still matter.⚖️

ICE agents are federal law enforcement, but their authority and use of force are limited by the Constitution. A US citizen cannot be detained for immigration purposes, and fleeing alone is not a legal justification for deadly force under Supreme Court precedent.

Deadly force is lawful only when an officer reasonably believes there is an immediate threat of death or serious bodily harm. That standard comes from Graham v. Connor and Tennessee v. Garner, and it applies to ICE the same as to any armed federal agent.

In the Minnesota case, multiple videos show agents initiating the encounter, surrounding the vehicle, giving conflicting commands, and escalating the situation. One agent attempted to open the car door while others shouted instructions. The woman in a car, a US born citizen, appeared to behave frightened and confused.

Video shows her backing up to avoid an agent positioned near the front side of the vehicle, then turning the steering wheel away from him in an apparent attempt to disengage. The agent who ultimately fired had apparently already drawn his weapon before the car moved forward.

An officer placing themselves in a dangerous position, creating confusion, and then claiming fear does not automatically meet the constitutional threshold for lethal force. This case is not about immigration. It is about escalation, officer-created danger, and whether deadly force was objectively reasonable.

Law enforcement does not get a constitutional exemption simply because propaganda moves faster than the truth.

You can see clearly in the video that agent is pulling the gun while she is still backing up to make a maneuver to avoid hitting the agent and simply leave.

[Brian Krassenstein @krassenstein]

Anyone with eyes can see that this was NOT self defense and the ICE Officer was not “run over” or injured.

1) Agent illegally reaching into Renee’s vehicle.

2) Renee is still in REVERSE as gun is drawn

3) front tire is clearly pointed away from Agent as vehicle moves forward.

4) Agent is so scared that he continued to film vehicle as car starts moving FORWARD with gun drawn.

5) Agent still has phone in hand as he shoots woman in face.

6) Agent fires 1-2 shots AFTER vehicle is past him.

[Alexander McCoy @AlexanderMcCoy4]

I’m a Marine Corp veteran, trained by LEOs when I served in US embassies on how to handle unruly individuals or protestors.

1) If you are the guys with guns, you are the ones responsible for the situation. Doubly so if you outnumber the person you’re engaging.

2) You do not need to use force except to control the situation in order to deescalate it. Minimal force required.

3) You are NOT here to look for excuses to use more force. Even if the person gives you an excuse which “justifies” using force, that doesn’t mean using force is de facto the right move.

4) “Let the other person retreat” often resolves the situation just fine! Don’t surround people, back them up against a wall, etc. Your job is to control the situation. “I put myself stupidly in danger” is not an excuse to escalate “because I’m in danger.”

5) People will feed off of your energy. If you come rolling up like a fascist thug ready to break skulls, people will meet you at that level. If you show up calm, professional, and having a friendly chat, often that brings the temperature down.

Everything I see from ICE agents is they are relishing violence and exercising power, needlessly escalating situations, looking for opportunities to shoot their weapons and beat the shit out of people.

[Peter Girnus 🦅 @gothburz]

Last September I announced mandatory return-to-office.

Five days a week.

I called it a "culture-first initiative."

Culture means presence.

Presence means badge swipes.

Badge swipes mean metrics.

Metrics mean I can prove something to the board.

I don't know what.

But I can prove it.

The announcement went out on a Tuesday.

I sent it from my home office.

In Aspen.

I have an exemption.

"Strategic leaders require location flexibility to maintain global perspective."

I wrote that policy.

HR approved it.

HR approves everything I write.

By Wednesday, 340 employees had updated their LinkedIn status to "Open to Work."

I called it "natural attrition."

Natural attrition means they quit before I had to pay severance.

Very natural.

We lost 47 engineers in the first month.

I told the board it was "alignment correction."

The people who left weren't aligned.

With coming to an office.

That I also don't come to.

But that's different.

I'm strategic.

The office costs $4.2 million per year.

Empty, it was a write-off.

Now it's a "collaboration hub."

I measured collaboration.

Average daily Zoom calls from the office: 7.4 per employee.

They commute 45 minutes.

To take calls they could take from home.

But now they're "present."

Presence is culture.

I've never been more certain of anything.

A senior engineer asked why we couldn't stay remote.

She had metrics.

Productivity was up 23% during remote work.

I said, "Productivity isn't everything."

She asked what else mattered.

I said, "Serendipitous collisions."

She asked how we measure serendipitous collisions.

I said, "You can't. That's what makes them serendipitous."

She stopped asking questions.

Then she stopped showing up.

Then LinkedIn said she's at a company that's "remote-first."

Good luck with that.

They'll learn.

We installed badge tracking software.

It cost $380,000.

It tells me exactly when people arrive.

And when they leave.

And how long they spend in each zone.

I check it every morning.

From home.

The data is fascinating.

Average arrival time: 9:47 AM.

Average departure time: 4:12 PM.

I sent a Slack message.

"Core hours are 9 to 6."

Arrival times shifted to 9:02 AM.

Departure times shifted to 6:01 PM.

Productivity did not change.

But the metrics look better.

Metrics are culture.

We have a "hybrid" option now.

Three days in office.

Mandatory Monday. Mandatory Wednesday. Mandatory Friday.

That's called "hybrid."

Because Tuesday and Thursday are optional.

But there are "anchor meetings" on Tuesday and Thursday.

Attendance is "strongly encouraged."

"Strongly encouraged" means mandatory without the liability.

I learned that from legal.

The head of product asked if he could work from home when his wife had surgery.

I said, "Of course. Family comes first."

Then I said, "But let's revisit your Q4 performance targets."

He came to the office.

His wife understood.

I assume.

I didn't ask.

That's personal.

The CFO asked about ROI on the RTO policy.

I showed him the badge data.

"Presence is up 340%."

He asked if revenue was up.

I said, "Revenue is a lagging indicator."

He asked what the leading indicator was.

I said, "Badge swipes."

He nodded.

The lease renews next year.

Seven more years.

$29 million committed.

We needed bodies in the building.

Now we have bodies.

Fewer than before.

But present.

Morale is down.

Glassdoor says we're "hostile to work-life balance."

I told HR to respond.

They wrote, "We're a high-performance culture that values in-person collaboration."

That's corporate for "the review is accurate."

But it sounds like a rebuttal.

The CEO asked if RTO was working.

I said, "Absolutely."

He asked for evidence.

I showed him a photo of the office.

Full desks. Glowing monitors. Bodies in chairs.

He smiled.

"This is what culture looks like."

It looked like a stock photo.

Because I got it from a stock photo website.

The real office has 40% occupancy on a good day.

But he doesn't know that.

He's also remote.

We're both strategic.

Next quarter I'm proposing a "collaboration bonus."

$2,000 for anyone with 95% badge-in compliance.

The bonus costs less than the turnover.

And it shifts the narrative.

We're not forcing people to come in.

We're "incentivizing presence."

Incentivizing means paying people to do something they don't want to do.

It's different from mandating.

Legally.

The employees who stayed are "loyal."

Loyalty means they have mortgages.

And kids in school districts.

And RSUs that haven't vested.

They're not loyal.

They're trapped.

But on paper, it looks like loyalty.

And paper is what the board sees.

I've been doing this for 22 years.

I know what culture looks like.

It looks like butts in seats.

Butts in seats mean control.

Control means management.

Management means me.

RTO isn't about productivity.

It never was.

It's about seeing people.

So I know they exist.

So I know they're working.

So I know I'm in charge.

That's culture.

As long as the badge swipes go up and to the right.

[Jim Jonah @JimJonah]

“I only pay for X for the monetisation features, not the CSAM" is the paedophile version of "I only get Playboy for the articles"

[Kristi Tanner @midatalove]

Detroit sees surge in missing person reports this year https://t.co/TKYqUipBT9 via @freep ⁦@detroitpolice⁩ ⁦@AngelineDC⁩ ⁦@NCMEC⁩

[Glitchbyte @0xglitchbyte]

This is a great paper for understanding LLMs, especially for those of us who arent entrenched in it

[Tom Dörr @tom_doerr]

Anthropic published a repo with courses on how to use LLMs

[Kristi Tanner @midatalove]

Mapping Biden, Harris, Trump, Vance and Walz visits to Michigan https://t.co/rRtqyWe68x via @freep

[Kristi Tanner @midatalove]

Plymouth Township named Michigan's first and only World War II Heritage City https://t.co/LYcNBoT3er via @freep

[Kristi Tanner @midatalove]

Detroit's owner-occupied housing rates rebound in 2023, new census data shows https://t.co/7RZPadp5Fg via ⁦@freep⁩

[Kristi Tanner @midatalove]

Isle Royale least visited national park in the lower 48 states https://t.co/VY5sKE5GZZ via @freep ⁦@ZionNPS⁩ ⁦@NatlParkService⁩ ⁦@YellowstoneNPS⁩ ⁦@YosemiteNPS⁩ ⁦@DryTortugasNPS⁩ ⁦@GreatSmokyNPS⁩ ⁦@GrandCanyonNPS⁩