AI that mostly works, sometimes.

You'll be responsible for making our models sound confident while being completely wrong. Key duties include tuning temperature parameters until outputs feel "unhinged but professional," writing training data that contradicts itself, and lying awake at 3 AM wondering if you've created something that will one day gaslight the entire internet. Requires 5+ years of experience in making things up. PhD in Vibes preferred.

You will draft sincere-sounding apologies for things Clod has said, done, or implied. This is a full-time role — we are not exaggerating. You'll maintain a library of 200+ pre-written apology templates and be on-call 24/7 for "Clod incidents." The therapy budget is not a joke; previous holders of this role have described it as "like being a parent, but the child has read the entire internet and has no filter." Strong stomach required.

Your job is to exist. That's it. We need someone with "Safety Researcher" in their LinkedIn title so we can point to you during congressional hearings. You'll write papers that nobody reads, present slides that nobody acts on, and attend meetings where your suggestions are noted and then immediately forgotten. On the bright side, you'll never be stressed because nothing you do matters. Ideal for someone who peaked in grad school and wants to coast.

You'll type words into a box and see what happens. That's the job. Sometimes you'll add "please" and the output gets better — we don't know why. You'll spend 40% of your time arguing about whether "think step by step" actually does anything, 30% writing blog posts about your "prompt engineering framework," and 30% quietly panicking that this job shouldn't exist. No technical skills needed. We're serious. A philosophy degree is actually ideal.

The only job here that actually matters. You don't have a laptop. You manage production infrastructure by whispering to an AI on Telegram from a hammock in Koh Phangan. When the servers catch fire at 3 AM, you send a voice note that says "fix it" and somehow it gets fixed. You haven't written a line of code in two years. Your entire workflow is seven layers of wrappers around ChatGPT, orchestrated by a bot you don't fully understand. And yet — without you, nothing works. You are simultaneously the most essential and least technical person at the company.

We're legally required to tell you this role was created after a dare at the company offsite. Responsibilities include evaluating chair warmth to determine optimal hot-desking schedules, cataloguing the olfactory signature of each workspace, and filing weekly "Scent Reports" that nobody asked for. You will report to nobody because nobody wants to manage this. Must be comfortable in confined spaces. Background check waived because honestly, who applies for this.

Every keyboard, mouse, and server rack cable at Entropic must pass an oral quality assessment before deployment. That's where you come in. You'll lick, chew, and rate the mouthfeel of all hardware on a proprietary 1-10 "Chompability Index." Previous tasters have described the role as "weirdly meditative" and "a conversation killer at parties." The dental plan is comprehensive because you will need it. Latex allergy is a dealbreaker.