kaechle

  Location: Brazil (US citizen)
  Remote: Preferably
  Willing to relocate: Never say never
  Technologies: Swift, macOS (XPC, SMAppService), MCP, Claude Code, Codex, TypeSafe Jev, LLM integrations (Anthropic, OpenAI), Python, asyncio, Pydantic, FastAPI, Postgres, Redis, TypeScript, React, Next.js, HTML/CSS, Tailwind, UI/UX design, technical writing
  Résumé/CV: https://www.linkedin.com/in/kaechle/
  Email: [email protected]

I'm Tim, an independent AI systems engineer, Swift/macOS developer and technical writer. Most of my recent work boils down to one question: what should a model actually get to see? Lately, that's meant building Scint, an agent workspace that takes a different tack on context. Instead of opening a thread and compacting it until it falls apart, you create a space. The space keeps its whole record, split by intent into nested transcripts, and a decision model (TypeSafe's Jev) routes each request so the agent gets only the slice of that record the work calls for. It's about 17,000 lines of Swift—a runtime, an MCP server, an XPC process daemon and a native macOS app—all designed and built by me.

The case study has the long version, with diagrams and real numbers: https://scintapp.github.io

I'm looking for work on LLM and agent systems: context engineering, decision-model routing, MCP integrations, or full-stack products with agents at their core. And if the role has room for the rest of my comb-shaped skillset (design, front end, writing), all the better.

I use both Claude and Codex for a lot of code- and prose-adjacent tasks, and I'm increasingly convinced that language models are quite terrible at, well, modeling. Not sure if that's ironic or not.

With the right blend of context and prompting, I can often get them to "lock into" an existing model, but using them to generate a novel outline or sketch, whether it's for an essay or a module, usually results in garbage.

You're arguing semantics; maybe missing the point entirely. The writer is using the analogy to emphasize personalization, not quality, or some other nebulous characteristic applicable to artisanship.

Correct. To decimate is the infinitive form of the verb decimate.

Using "literally" figuratively or, more precisely, as a hyperbolic intensifier [0], is a tradition employed by notable English writers who lived and died long before you were born.

[0] https://en.wikipedia.org/wiki/Literally

Those are some bounding leaps you made without much context. Are you in sales?

Kidding aside, my first reaction was: perhaps the occasions they were aware of their own influence were ones in which they didn't much care for the outcome. Or maybe a conflict of interest, like trying to win over a hiring manager for a position you know you'll hate.

I don't think cajoling or persuading others inherently manipulative, but I can think of a lot of examples where doing so feels grimy.

I’ll see your pedantry and raise you ... more pedantry. The sentence may be a bit clunky, but there’s nothing grammatically wrong with it. And you’re leaving out the first sentence, which frames the comparison:

> ASML gross revenue was 28B€ in 2024, and their net income was 7.5B€. While 1.3B€ (the amount ASML invested in this 1.7B€ fund raise) is not pocket change, it is also an amount that ASML can not afford to lose.

Worded another way:

> ASML had a healthy margin of 7.5B€ on 28B€ in gross revenue in 2024. 1.3B€ isn’t a huge chunk of this, relatively speaking, but *it’s also an amount that ASML can’t afford to lose.*

Still clunky. Still not wrong.

SEEKING WORK | Remote Preferred | US/Brazil

  Technologies: Python, asyncio, Pydantic, LLM integrations (OpenAI, Anthropic), Meilisearch, WebSockets, Redis, Postgres, REST/HTTP APIs, Front-end web development, React, Next.js, HTML/CSS, Tailwind, UI/UX, design, technical & content writing
  Résumé/CV: https://www.linkedin.com/in/kaechle/
  Email: [email protected]

I’m Tim, a developer and writer with a swath of experience across the full stack, from design to Python-based backend systems, including asynchronous operations, concurrency, and integrated LLM workflows. My recent work involves orchestrating and managing processes with libraries like Meilisearch, OpenAI/Anthropic endpoints, and advanced metaclass/typing/patterns in Python. I also have a strong front-end background, keen design sensibilities, and proven writing and technical documentation skills—my writing portfolio can be viewed on my LinkedIn profile. I’m currently looking for work that challenges me technically and allows me to communicate complex ideas through writing and design.

Location: Brazil (US Expat)

Remote: Preferably

Willing to relocate: Never say never

Technologies: Python, asyncio, Pydantic, LLM integrations (OpenAI, Anthropic), Meilisearch, WebSockets, Redis, REST/HTTP APIs, Front-end web development, React, Next.js, HTML/CSS, Tailwind, UI/UX, design, technical & content writing

Résumé/CV: [LinkedIn Profile](https://www.linkedin.com/in/kaechle/)

Email: [email protected]

I’m Tim, a developer and writer with a swath of experience across the full stack, from design to Python-based backend systems, including asynchronous operations, concurrency, and integrated LLM workflows. My recent work involves orchestrating and managing processes with libraries like Meilisearch, OpenAI/Anthropic endpoints, and advanced metaclass/typing/patterns in Python. I also have a strong front-end background, keen design sensibilities, and proven writing and documentation capabilities—my writing portfolio can be viewed on my LinkedIn profile.

I’m currently looking for a place where I can continue learning and growing my broad skillset; a position where I can challenge my technical skills as well as my ability to communicate complex ideas through writing and design.

This is the way. Blending two or more styles also works well, especially if they're on opposite poles, e.g. "write like the imaginary lovechild of Cormac McCarthy and Ernest Hemingway."

Also, wouldn't angry Charles Bukowski just be ... Charles Bukowski?

Location: Brazil (American expat, legal address in Las Vegas)

Remote: Preferred

Willing to relocate: For the right opportunity

Technologies: Python, FastAPI, Pydantic, Asyncio, JavaScript, TypeScript, Node.js, Next.js, HTMX, Tailwind CSS, PostgreSQL, Redis, API design, LLM API integration, Docker, Git, REST APIs, React, SQLAlchemy, Pytest

Résumé/CV: https://www.linkedin.com/in/kaechle/

Email: [email protected]

For the past year and a half, my passion project has been developing a comprehensive Python framework designed for building modular, asynchronous, and event-driven LLM applications. It features a component-based architecture encompassing state and context management, AI-composable workflows, and code parsing and analysis. The goal is to create an intelligent system with a persistent context/memory layer built dynamically via search while providing a natural-language interface for workflow composition. I'm eager to explore AI engineering opportunities that align with this trajectory.

Great points. We're pattern-matching shortcut machines, without a doubt. In most contexts, not even good ones.

> When I go to the doctor, I don't study medicine first, I trust the doctor. Trust takes the place of genuine understanding.

The ultimate abstraction! Trust is highly irrational by definition. But we do it all day every day, lest we be classified as psychologically unfit for society. Which is to say, mental health is predicated on a not-insignificant amount of rationalizations and self-deceptions. Hallucinations, even.

Every time I read "stochastic parrot," my always-deterministic human brain surfaces this quote:

> “Most people are other people. Their thoughts are someone else's opinions, their lives a mimicry, their passions a quotation.”

- Oscar Wilde, a great ape with a pen

This. Em dashes are—as punctuation goes—downright sexy, so I committed shift+opt+- to muscle memory long ago. And it's arguably less work than a bucked-toothed double hyphen.

Btw, if you haven't already, check out Karabiner on macOS for AutoHotkey-esque functionality.

:clown + :pointright + ?!

(Surely, you jest?!)

  Location: American living in Espirito Santo, Brazil (with residency)
  Remote: Preferably
  Willing to relocate: Never say never
  Technologies: HTML, CSS, Tailwind, Svelte, JS, Python, Node, SQL, Linux, nguni, Docker
  Résumé/CV: https://www.linkedin.com/in/kaechle/
  Email: [email protected]

I have an array of skills that span way too many disciplines, including web and ui design, web development, technical content writing, database admin, and a whole lot more. Lately I've been building custom AI tools using Python, and I'd love to continue down this path, but I'm open to positions that can fully leverage my comb-shaped skillset.

Trite and rehashed groupthink—marketing garbage at its finest. But you should still write. As should every other human being, vocational identity notwithstanding.

> Sums up 94.7% of the internet.

Fixed.

I'm still in the early stages of learning how to code, but I've started doing something similar to help me get into the right mental state. As someone with severe ADHD, it takes serious effort to get the brain waves waviating at the right frequencies. If I mentally check out, even for a couple of hours, it's like I completely forget everything I've worked on.

Out of frustration, I started talking through my code out loud, and the difference was incredible. All my mental scaffolding and abstractions sprang to life, where before it would've taken an hour or two of rereading my code and rebuilding everything in my mind. Now I'm a chatty bird, typing and mumbling to myself as I work.

Also, not to discount your struggle, but don't write yourself off as a bad communicator—pun intended. I spent 15 years in tech, then seven years writing marketing content, and now I'm transitioning back to tech. Words are hard. The ability to string them together coherently does not make someone a great communicator. Like writing code and engineering, it's easy for laymen to conflate the skill with the vocation.

Keep practicing—you'll eventually outshine those who think opening their mouths and making noises is necessarily good communication. And check out some famous speakers who're on the spectrum if you haven't—you'd be surprised.

The test states at the end that dogmatism is measured based on the answers to a subset of the questions—the game is simply a correlate.

Likewise interested and exploring the same area.