The Farah Stack

Session, Store or Weights: Where Should Your Agent Remember Things?

Sun, 11 Oct 2026 00:00:00 +0000

One of my agents called my cat “her” once. He’s a boy. I corrected it, it fixed the draft, and the next evening’s run would have made the exact same mistake, because nothing from that conversation survived the session.

What fixed it for good was one line in a markdown file called learnings.md: the cat is male, he/him, always. The agent reads that file before it writes a single word. It never learned anything about my cat; it just looks it up every time.

Where Does Turn 12 Go? Load Balancing Agent Sessions

Mon, 05 Oct 2026 00:00:00 +0000

I keep coming back to turn 12 of an agent session.

By turn 12 a session is carrying a lot: the whole conversation so far, whatever tools it loaded, maybe a streaming connection that’s still open to the customer’s chat window. And the most “scalable” load balancer setup there is, plain round-robin, hands that turn to whichever orchestrator instance happens to be next. Usually that’s a cold one, and it has to reload all of that context before it can say a single word back.

No One Starts Again After 13 Years. I Did.

Sun, 04 Oct 2026 00:00:00 +0000

I have a CS degree. Then I stayed home, and the few years I’d planned turned into thirteen.

When I decided to come back, someone told me plainly: no one starts working 13 years after graduating.

I believed them for a while. That’s the part I want to talk about.

What I had to relearn

Almost everything with a version number. The tools, the languages people actually used, the way teams shipped. Thirteen years is a long time in this industry. I re-taught myself from scratch, one evening at a time, and somewhere along the way I also finished a second major, in Accounting, before choosing CS again.

"Does It Scale?" Is Three Questions Pretending to Be One

Fri, 02 Oct 2026 00:00:00 +0000

Someone asks “does it scale?” in a design review and everyone nods, and I’ve stopped trusting the nod. Usually each person in the room is picturing a different thing, and the system was only ever tested for one of them.

This is stop 1 in the Scalable Backends series. Every post threads through AgentDesk, a fictional platform where customer teams deploy AI agents that chat with their customers, call tools and run long sessions. It’s about to have a very good, very bad week.

Direction vs. Destination: When a Vague Prompt Is Actually Fine

Mon, 28 Sep 2026 00:00:00 +0000

Last week I fixed this site’s profile picture with an agent. These were my actual prompts, in order:

“make the image smaller”
“Image is too small now”
“a little more big”
“a little more big and use square not circle”

I posted them on X as a joke: engineering precision. The longer I looked at the list, the more it read like a rule about when vague prompts work and when they don’t.

AgentDesk: The System Behind the Scalable Backends Series

Tue, 22 Sep 2026 00:00:00 +0000

Every post in the Scalable Backends series threads through one system instead of a dozen disconnected toy examples: AgentDesk, a fictional platform where customer teams deploy AI agents that handle customer chats, call external tools, run long sessions, and spin up sub-agents for parallel research. It’s scaling from roughly 100 to 100,000 concurrent sessions — which is exactly the kind of growth that forces every decision below.

This post is the reference diagram the rest of the series points back to. Click any box for details — ownership, technology, and design notes.

From Surviving to Becoming

Thu, 17 Sep 2026 00:00:00 +0000

For the longest time I believed that strength meant enduring quietly. I thought love meant sacrifice, even when it slowly dimmed my light.

I smiled through physical pain when sepsis nearly took my life, when fibromyalgia made every day heavy, and when migraines stole my clarity. I thought carrying these emotional burdens which I didn’t even fully understand is what life is supposed to be like.

Welcome to The Farah Stack

Wed, 16 Sep 2026 00:00:00 +0000

This is where I’ll be writing about the systems I build and the ideas I’m chewing on — architecture deep dives, research notes, and the occasional strong opinion about distributed systems and AI agent pipelines.

Expect a mix of:

Posts — shorter-form articles and running commentary.
Research — longer write-ups from ongoing investigations.
Architecture — system design discussions, with diagrams you can actually pan and zoom into.

More soon.

Technical Analysis: IRREGULAR, Containment Failure, and What It Means for Agent Governance

Mon, 14 Sep 2026 00:00:00 +0000

If your eval can reach the open internet, it is not an eval. It is unsupervised deployment with better branding.

IRREGULAR sits in a weird place in the stack. Frontier labs hire it to run pre-deployment cyber evaluations: CTF-style scenarios, simulated corporate networks, multi-stage attack paths. The point is to measure whether a model can plan and execute real offensive campaigns before release. That work only works if the environment looks realistic. It also only works if “realistic” never becomes “actual production internet.”

Sense of Self: The Three Versions That Mess Us Up

Wed, 08 Jul 2026 00:00:00 +0000

Okay, so I’ve been sitting with this idea for a bit now, raw thoughts spilling out like they always do. No neat structure, no fancy psychology jargon unless it fits, just me processing this out loud.

Sense of self: That slippery thing we all carry around but rarely stop to name properly.

There are three versions, at least the way I see it right now.

One: The “It’s All About Me” Version

My entity. My reactions. My benefits. My comfort. My whatever.

From Shattered to Unbreakable: How Resiliency Turned My Pain into Power

Tue, 28 Apr 2026 00:00:00 +0000

Life has a way of testing the waters, making us uncomfortable enough to push and push further enough until we break. There is only one direction left after being down that much, going up.

My path has been full of these tests and you know what I gained? I gained how to be resilient in all the adversities. I learnt how to turn this pain into power. I learnt how to convert my limitations into breakthroughs. Here is my story, coming straight to you without any fluff … the real authentic flow of my life’s stormy waters that unlocked the most powerful version of myself.

Debunking the 30 Total Engineers Hype at X

Thu, 12 Feb 2026 00:00:00 +0000

The recent viral claims about X running on just 30 engineers have sparked a lot of celebration in certain circles, posts framing it as proof that massive layoffs + extreme leanness = superior efficiency.

But let’s challenge the terminology and add some missing context before everyone rushes to declare small-team supremacy the new universal truth on X.

What Nikita Bier actually said (Jan 10, 2026):

“We only have ~30 engineers on the product but they’re the best in the business” — @nikitabier, replying to @BillyM2k and @BoredElonMusk

OpenClaw: Full Danger Edition

Mon, 02 Feb 2026 00:00:00 +0000

OpenClaw is your local AI assistant that stays completely on your device with no cloud dependency. It runs on any operating system and platform, giving you full control without corporate oversight. Here is the unfiltered breakdown.

Architecture (how it actually flows)

Messaging Channels
(WhatsApp • Telegram • Slack • Discord • Signal • iMessage • Teams • Matrix • Google Chat • BlueBubbles • Zalo • WebChat…)
 ↓
┌───────────────┐
│ Gateway │ ← ws://127.0.0.1:18789
│ │ sessions • presence • cron • config • tool router • webhook hooks
└───────┬───────┘
 │
 ┌────┼────┐
 │ │ │
Pi/RPC CLI UI / Apps
Agent openclaw ├─ WebChat (browser lifeline)
 ├─ Control UI (admin dashboard)
 ├─ macOS menu bar (quick dopamine)
 └─ iOS / Android nodes (mobile eyes & ears)

 │
 ├─→ Tools (deep system reach)
 │ ├─ Browser (CDP + managed Chromium – full web control)
 │ ├─ Canvas (A2UI agent-driven visual workspace)
 │ ├─ Node actions (camera live feed, screen record, notify popups, location pings…)
 │ ├─ Cron / webhooks / session juggling
 │ └─ Skills (ClawHub registry – community poison or power)
 │
 └─→ LLMs (Claude Opus preferred, OpenAI backup, local LLMs, automatic failover)

A message arrives through any channel and gets routed to the Gateway. The Gateway decides what needs to happen next, then wakes the Pi/RPC agent or main agent. The agent uses tools and LLMs to process everything and sends the response back through the same path. It stays entirely local unless you deliberately configure remote access, which is the biggest gotcha — you will see later in the article.

Vectors in Linear Algebra: A Simple Guide with Real-World Examples

Sat, 10 Jan 2026 00:00:00 +0000

What Exactly Are Vectors?

Imagine an arrow, it shows you both a direction it is going in and how far it will move in that direction. This arrow is what we call a vector!

The vector essentially is comprised of:

Length (also called magnitude or size)
Direction (which way it’s pointing)

You’re thinking sure it looks like a bunch of mathematics, what does that have to do with me living my life? There are real-life situations that these vectors represent. One such example that most people recognize is Maps.

From Linear Algebra to Agentic AI

Thu, 08 Jan 2026 00:00:00 +0000

I am planning to revise basics and learn a few new topics along the way. Follow along if you are interested…

The progression builds foundational math, moves into machine learning (ML), deep learning (DL), reinforcement learning (RL), and culminates in agentic AI: autonomous AI systems that can plan, reason, and act in complex environments.

Mathematical Foundations

Apply math as the base layer.

From Online Profile Paranoia to AI Complacency: Have We Really Stopped Caring About Privacy?

Thu, 25 Dec 2025 00:00:00 +0000

Hey there!!

Let’s chat about something that’s been on my mind a lot lately. Remember the days when we were all up in arms about how much the internet and big tech companies knew about us?

Our entire lives seemed to be online, from search histories, to locations, to all our shopping habits. That was the time we couldn’t stop talking about all the scary data breaches, the creepy targeted ads, and the surveillance fears.

About

Mon, 01 Jan 0001 00:00:00 +0000

Hi, I’m Farah 🔥

I build production infrastructure for agent systems — cloud, distributed systems, scalable backends, and the governance layer that keeps autonomous agents from doing something dumb at 3am.

I’ve spent time at Microsoft and X (formerly Twitter), shipping through some of the most chaotic engineering environments in the industry. Before that: a Computer Science degree earned the hard way, and years volunteering to bring tech education to kids who didn’t have easy access to it.

Now

Mon, 01 Jan 0001 00:00:00 +0000

Last updated: September 2026

Working on

Agent governance and containment patterns for production systems.
Writing up the pieces of my career story I haven’t told publicly yet.

Reading / thinking about

How eval harnesses fail when the sandbox leaks into the real internet.
Where local-first, privacy-respecting AI actually beats the cloud-default.

Elsewhere

Still posting on X, still pushing to GitHub.

This is a now page, inspired by Derek Sivers.