Remote: Yes

Willing to relocate: Yes

Technologies: TypeScript, JavaScript, Python, PHP, Rust, React, React Native, Svelte, Express, Bun, Angular, GraphQL, D3.js, visx, Backbone, jQuery, LangChain, Mastra, FastAPI, pandas, scikit-learn, Optuna, Chrome Extension API, Playwright, Electron, Stagehand, browser-use, Web Audio API, WebRTC, WebSockets, PostgreSQL, TimescaleDB, MySQL, MongoDB, Redis, AWS, EC2, S3, Lambda, Docker, Git, LLM agent design, agent evaluation, reinforcement learning, browser automation, MCP

Résumé/CV: N/A

Email: [HN username]@gmail.com

For one year I reverse-engineered the major education platforms — Google Docs, Microsoft Office, Nearpod, Kahoot, ect. — so that AI agents can drive them on a teacher's behalf and automate classroom workflows. The work runs in over 10,000 classrooms today.

Building agents against production code you don't control is harder than it sounds. The code is mangled, so first you have to understand how the app was built and then take it apart from the outside. It's reverse engineering, but it depends on knowing how to build the thing in the first place.

The role came with monthly compliance and security training centered on FERPA, so I'm fluent in handling student data responsibly. That's not new ground for me. I spent five years building on Drupal, which the most prestigious learning institutions still run today, and version control and access control were the hard part of every project. Student data is the same problem.

I've been building browser agents and browser automation since 2018, well before it was a category. That's paired with 13 years of building dynamic, data-heavy UIs for web and mobile.

I've also built admin dashboards for a custom legal document management system. And a CRM for email marketing, with a drag-and-drop interface for designing and branding email templates for social campaigns. I've taken several companies from 0 to 1, as a consultant and as a full-time employee, across fintech, streaming, real estate, education, marketing, and media. I've worked at a 130-person AI company and on a 7-person team where I wore every hat. I do my best work on small, fast-moving teams.

https://github.com/adam-s

https://adamsohn.com

I was thinking the other day how much better Drupal is. Want a online store? A few commands and bam, online store. Want a newspaper? A few commands and bam, newspaper with publishing workflows, user management, and caching.

Using coding agents isn't much different. There are several things the models are trained to do very well and a few commands will get something. If the developer wants to move the project beyond that, it requires domain knowledge and a lot of hacking.

I wonder if the coding agents will move towards the Drupal model where they create interchangeable components with common interfaces. Like Drupal the coding agents never provide anything truly inovative that hasn't been done before.

I have a very niche set of skills so I could up until 6 months ago pick up contract work anytime I needed. Despite being one of the best in the world at what I did, I can't compete anyone with $400 in tokens using Codex or Claude Code. I'm pivoting quickly but the sentiment is "Oh, shit, this is coming fast and heavy!"

Word of advice.

Get a part time job where you can keep your skills up-to-date because from personal experience personal projects on GitHub will not be enough to land a job in two years. Make sure for every penny you spend you are earning equal.

The two years are going to fly by.

EDIT: I'm not joking about taking this serious. You want to be working when people start losing their jobs to AI. Most likely this is going to end with society rethinking distribution but you are going to need to be able to survive the changes and 2 - 3 years is not enough wealth.

They are dumping them on your 401k -- especially SpaceX.

Diana Krall opened and by open I mean she was the first of 4 acts and had a free concert a couple days before at a coffee shop down the street — practically unknown.

Sonny Rollins headlined and for the encore he played La Cucaracha. After about 20 minutes of La Cucaracha the pianist was signaling to the sound engineer to cut the sound. 5 minutes later the band one by one started to put down their instruments walking off the stage. Sonny Rollins kept belting La Cucaracha from his saxophone probably for another half hour or more after that. Life is good.

If I get desperate, I might find the the cut off for 200 - 300 companies and try to honestly get my resume into those parameters. At this time, it is below me to do that.

The job websites are requiring a linkedin profile in order to submit an application. I had one interview and the guy sounded like a linkedin stream. So I thought about updating mine. I was immediately hit with political propaganda. From the point of view of an established company, it is a test that the person is willing to conform to social standards. It is the modern equivalent of wearing a tie to an interview.

I miss the good ol' days when they would look at my resume, call the places to make sure I worked there, and invite me in to prove I know the skills that I put on my resume.

Yet, the biggest bullshittery, is every company that almost each of you work at requires a link to a LinkedIn account on every job application, not optional. It has become a form of social credit. LinkedIn isn't completely meaningless either. A huge portion of the posts are also propaganda. Finding a new job is tied to listening to propaganda.

The can't predict the consequences of an action predicting one token after another. They can't solve a Rubik's Cube unlike a 7 year old human who can learn to do it in a weekend. They can't imagine the perspective of being a human being unlike a 7 year old human if asked to imagine they where in the position of another human.

Do you want a demo of what this is capable of?

If we shift the paradigm of how we approach a coding problem, the coding agents can close that gap. Ten years ago every 10 or 15 minutes I would stop coding and start refactoring, testing, and analyzing making sure everything is perfect before proceeding because a bug will corrupt any downstream code. The coding agents don't and can't do this. They keep that bug or malformed architecture as they continue.

The instinct is to get the coding agents to stop at these points. However, that is impossible for several reasons. Instead, because it is very cheap, we should find the first place the agent made a mistake and update the prompt. Instead of fixing it, delete all the code (because it is very cheap), and run from the top. Continue this iteration process until the prompt yields the perfect code.

Ah, but you say, that is a lot of work done by a human! That is the whole point. The humans are still needed. The process using the tool like this yields 10x speed at writing code.

I'm stating that LLM models are not capable of predicting the consequences of their actions which makes in inept with spacial and temporal understanding of the environment state.

I like the Rubik's Cube because it is a harness that helps me try to develop a prompt to get reasoning models to reason about the consequence of an action.

[0] https://mastra.ai/

1. a human child learning 6 algorithms and a weekend can solve a Rubik's Cube

2. Reenforcement learning can solve a Rubik's Cube

3. The best LLM model using recursive tuning or not can't solve a Rubik's Cube.

Claude 4.6 got 60% of the way but couldn't figure out the last steps after running for 20+ minutes and hundreds of thousands of tokens.

Comments URL: https://news.ycombinator.com/item?id=47977959

Points: 4

# Comments: 0