Search Results

7 results

Peter Vijeh

https://www.petervijeh.com/

Peter Vijeh — platform engineer in New York. I build the CI, identity, and agent infrastructure that lets teams ship with AI coding agents without lowering the bar.

I scanned 1,074 companies for a SaaS worth cloning. The best answer was “don’t.” — Peter Vijeh

https://www.petervijeh.com/projects/saas-scanner

A multi-source scan of Constellation Software's entire portfolio, scored for solo-founder attackability and then re-screened on a distribution constraint: can a stranger buy it without a sales call? 1% publish a price — 2–3% once you correct for the crawler only reading English. The seed list was selected against me, and the negative result is the deliverable.

AI is supposed to be deflationary. Here's what the macro data suggests. — Peter Vijeh

https://www.petervijeh.com/projects/ai-deflation

Is AI deflationary? Zillow, FAO, EIA, 32 software 10-Ks, five status pages and BLS productivity tables, no CPI. Household prices rose after 2022, software operating margins went from 29% to 34%, status-page incidents roughly doubled, and token prices fell 80 to 90%.

Does Reddit have an astroturfing problem? What the data suggests — Peter Vijeh

https://www.petervijeh.com/projects/reddit-astroturf

In 51,129 comments from six knife subreddits, the 5% most brand-heavy accounts wrote 11.3% of the brand mentions in buying threads, against 7.9% expected by chance. For one brand it is 31% against 8%. The full Reddit histories of those accounts are 4.5 years old and spread over 66 subreddits, the same as a comparison group, so the data shows concentration and not who paid for it.

I had Gemini train its own replacement for $9 — Peter Vijeh

https://www.petervijeh.com/projects/reddit-ner

Gemini 3.1 Pro labeled 4,290 Reddit comments for $9; a fine-tuned GLiNER model now tags brands, models and materials locally at 0.83 F1 against those labels. Five of ten training runs produced nothing because of a tensor I misread as an attention mask; the loss went flat and no error named the tensor.

Gentrification signals are a lagging indicator of home prices, not a leading one — Peter Vijeh

https://www.petervijeh.com/projects/zip-forecast

Fort Greene went from 31% below the New York metro median home value in 2000 to 86% above in 2018; its coffee shops arrived after 2012. Across 782 metro ZIPs, coffee-shop growth tracks the price change that ended four years earlier, the 2005-2015 winners lost in 2015-2025 (rank correlation -0.62), and whether cheap ZIPs catch up flips sign every six to eight years. A model trained on 26 years of Zillow and Census data ranks next year's ZIPs at +0.44; sorting by last year's growth gets +0.39; at ...

Is Manhattan pricing out the young? What 14 years of data show — Peter Vijeh

https://www.petervijeh.com/projects/nyc-movers

I set out to prove that fewer young people, and fewer broke young people, move to New York than in 2011. Three federal datasets later: the survey has arrivals aged 20 to 29 flat citywide and down 15% in Manhattan from 2011 to 2024; the IRS has both down about 20% from 2011 to 2020, then stops being comparable. Arrivals earning under $50,000 in 2024 dollars are down 29% in Manhattan, and the people who never moved got richer at almost the same rate.

Filters
Category
Domain
In web feed
Language