Pre-2023 Facebook
In the hazy before-times of 2014 tech, I joined Facebook. I was the first person hired specifically as an engineer in an Austin office that employed fewer than 200 people in all roles. One of them was my brother. It was not a company back then so much as a cult. My first 6 years there were the hardest I’ve ever worked, the most I’ve ever learned, and, to quote a project manager of mine at the time, a ticket into “a theme park for psychos,” who would gleefully stay after work to keep talking about work.. we just also added drinking and played video games when 6-7pm or so rolled around.
For ~6 months I was a completely useless little ball of stress held together by imposter syndrome, intense curiousity, and a stubbern refusal to admit defeat by a serious of 1s and 0s. Then, all in one moment, I woke as an “SME” - someone who other people went to for advice and help and was considered senior, and In Charge Of Things. This was not because I all of a sudden had become an expert; it was just a common on a team where we had only 1 engineer for every 300 million users, and if you’d been around for 6 months, that was longer than 50% of everyone else. But it didn’t help with the imposter syndrome.
For about 8 years after than, my job was roughly divided like this:
- 15% tackling really, excellently fun, important, and hard problems that I loved, even as they burned out my brain.ZZZ
- 30-40% answering the questions of whoever came by my desk
- 95% providing therapy for the totally valid and stress-induced nervous breakdowns of my coworkers in small 1:1 conference rooms
- 100% tech-driven pranks and meme wars with my team, who, predictably, after eating all of our meals together, working all day, and spending time together after work, came up with some pretty intricate ones.
Finding Focus with What 3 Things
After a few years of finding myself accidentally working as a mostly full-time therapist in addition to my day job, I kicked off a weekly post called “What 3 Things Alyshia’s Doing This Week.” I had two main strategic goals for it:
- Keeping Myself Focused: Writing these posts helped me prioritize and avoid saying “yes” to too many things at once.
- Scaling Therapy: I hoped it would encourage my newer engineers (who I always called younglings, regardless of age) to take a deep breath and realize it’s okay to not finish everything all at once.
Those posts became a ritual over time that helped me to incorporate balance, calmness, and perspective in an evironment where the potential for error was huge, and everything always seemed to be on fire.
When Everything Must Be Done at Once
Sometimes, though, the idea of “not doing everything at once” just doesn’t hold up. Right now, I’m in one of those moments where I do need to tackle everything at once. So, I’m restarting my “What 3 Things Alyshia’s Doing This Week” posts with a clear purpose:
- Track Eeverything: By documenting my priorities, I can monitor the size of my workload over time.
- Ease the Crunch: My goal is to watch that workload shrink as I chip away at it before the September 12th deadline.
I can’t get this down to 3 lines yet, but the goal is to get there by October. Here’s to managing the madness, finding balance, and coming out stronger on the other side!
#TODO Silly List 📝
Make a TODO list✅ (Finally accomplished something today!)- Actually figure out what all of my projects are, what domain manager
- Document domain managers & where they are stored
- Write a proper README.mds that doesn’t just say “it works on my machine” for my 50 github repos (this iw now fully automated!)
- Add the projects that have been living on my local for btw 1-12 years, add them to github or bitbucket
- Document installation steps (beyond “here’s a curl command”)
- Explain why I chose this tech stack (spoiler: I didn’t, my m2 chip chose for me)
- Add usage examples make sense, to actual humans
- Write contributing guidelines (move fast, break things, plz don’t break everything)
- Create a proper license file (currently using the “please don’t sue me” license; need the MIT one like the rest of the cool kids)
- Add badges to make it look professional (even though it’s held together with duct tape)
- Document the 47 environment variables needed to run projects scattered in various .yml, Gemfile, package.json, and [insert_pkg_name_here].config
- Explain the mysterious
config.json.example.backup.old.final.FINAL.swp
file - Add screenshots (real ones)
- Write changelog (v1.0: “it exists”, v1.1: “it still exists”, v2: “it’s broken” v2.1 “fixed it” v2.2 “-> profit”)
- Fix that bug that happens only on Tuesdays during a full moon
- Fix jekyll build bug that appears to only happen in Vercel :/
- Add tests (currently using the “if it compiles, ship it” methodology)
- Nvm, I just remembered I’m OG FB, #testinprod
TODO List, for real though
- Finish good-enough homage clone styling for homage website ahead of the 31st
- Write August 31st Homage post
- Post August 31st Homage post
- Update this post with 2025 data from this website
- Update this post with data from my own github profile
- Use this post and this post to explain why a neural network trained on your brain would be better than you are at golf, or knowing your own name.
- Read this article on NN performance, which was probably out of date the moment it was published.
- Work these two quotes somewhere into this website:
- “Selection bias is a hell of a drug.” ~F. Perry Wilson, MD, MSCE
- The price of reliability is the pursuit of the utmost simplicity. ~C.A.R. Hoare (1980 ACM Turing Award Lecture)
- Contribute to the ML Commons AI Risk & Reliability working group
- Integrate BrightData into my SingleSiteScraper to scale it and prettyify it
- Figure out if I want to use AWS SageMaker
- Submit and find more silly competitions like this
- Create MCP out of HHSC compliance research that spits out a TODO-list
- Create MCP out of all Integrity Studio’s data and turns it into even more Todo-lists
- Create MCP for my various important filesystems
- Self-host n8n, probably on Docker
- Look into this use case for BrightData as a quick-start
- Create some truly ridiculous stuff with my MailSlurp MCP server
- Download some AmazonQ rules to my local
- Follow [this format] for integrating schema.org and json data
- MCP server that auto-identifies appropriate schemas out of html code
- Read OpenAI’s MCP Server specs and make one with it
- Load this as a best-practices README.md into core amazonq resources
- Parse this very comprehensive list of json struct definitions
- Play around with these typescript data tools
- Another data schema to play around with/make an MCP server out of
- MCP server that scrapes all of my emails, docs, messages, regular websites, and creates yet more TODO lists and emails out of them - probably using Bright Data
- FINISH COMPLIANCE RESEARCH AGENT; mvp complete
- FINISH Dance calendar scraper as example for calendar scrapers & importers in general