tldr; I’m starting a new role, but the scraping+recommendations are available at occurai.com

recruitment was taking too much time, and it was initially planned only as a source for data and revenue for AI research, so I tried to focus on the research in the last year [geomtetric DL, adjusting topnets from simplicial complex compute to spectral sequence aka from homology to homotopy], but faced some hardware limitations, and during market research tried to join either research groups in geoDL, or new HW compute architectures, the latter seemed more promisable and I’m joining one of them that utilizes superconductivity

the current platform functionality at occurai.com:
everything optimized to low-cost [currently I pay no more than $100/month for cloud], so the solution is simplified, no LLMs used [but I use transformer model for embeddings]
- I source companies manually [currently ~1400 companies, brief you can see at ‘jobs summary’ page, placeholder is for either remote or hybrid or not_found/ambiguous location]
- every 30 days scraper updates the current openings from the career pages
- recommendations are recomputed every 24h [appr near neighbor + some transformer, used only jobs in the last 30 days]
- currently there’s no hard match on location
- subscription is $10/month, just to cover cloud costs [will either go down with users, or I’ll add compute/llm for a better match and sourcing], but feel free to skip and just use LI, or YC, or remote etc.

how did I find the job:
- leetcode
- hardcore market research given my background
- tuned up resume [didn’t lie in it, but stretched to the max]
- applied to ~1k jobs in the last year
- talked to ~50, most were agentic startups, but noticable are Shopify and Deepmind
- failed a few tech interviews
- emails to the companies that have no open roles yet [that’s how I got the offer eventually]