17 months of Hacker News job posts show a modern stack shift

A public analysis of 7,066 job postings from Hacker News “Ask HN: Who is hiring?” threads and five other job boards says the current tech hiring market is centered on a fairly specific stack: Python, TypeScript, React, AWS, PostgreSQL, and an expanding set of AI/ML terms.

The report, published by the HN user behind the scrape, covers 17 months of postings from January 2025 through May 2026. It filters the full set down to 6,539 tech-relevant jobs and tracks 85 skills, 30 common fluff phrases, experience requirements, and explicit salary ranges.

According to the dataset, Python was the most frequently mentioned skill in the tech subset. But the report also says TypeScript plus React together describe a frontend stack that appears in about one in four postings, putting frontend demand at a similar scale to Python.

On the language front, the report says Python led TypeScript by seven percentage points, with Rust overtaking Java in the Hacker News-style hiring pool. React appeared in 22.0% of tech postings, while Vue, Angular, and Svelte combined accounted for less than half of that share.

The frontend findings also point to a shift in framework preferences. Next.js appeared in 5.5% of postings, which the report says means about one in four React jobs explicitly mentioned Next.js. Tailwind showed up in twice as many postings as it did a year earlier, while the report says the CSS-in-JS era appears to be fading at the trend level.

Cloud hiring was concentrated as well. AWS appeared in 15.0% of postings, compared with 5.9% for Google Cloud and 3.0% for Azure, a roughly 3:1 ratio of AWS to GCP and 5:1 AWS to Azure in the sample. The report says multi-cloud language was rare in the job descriptions it analyzed.

The data layer was similarly lopsided. PostgreSQL appeared in 15.0% of postings, on par with AWS, while MySQL “barely registers,” according to the report. Redis dominated caching mentions, and MongoDB still appeared but less often than before.

AI-related hiring was a major part of the sample. The report says LLM appeared in 11.5% of all tech postings, meaning roughly one in nine tech jobs explicitly touched large language models. Anthropic mentions slightly beat OpenAI mentions in the dataset, at 3.24% versus 2.86%.

The report also breaks out model frameworks. PyTorch appeared in 133 postings, while TensorFlow appeared in 39, a 3.4x gap in favor of PyTorch. In the most recent three-month window, TensorFlow was the fastest-trending framework, up 120% versus the previous three months, though the report cautions that this could reflect sampling noise.

Experience requirements were another notable finding. Of the 418 postings with an explicit minimum years-of-experience requirement, the biggest bucket was 3 to 5 years. The report says only about 7% asked for 10 or more years.

The “Senior” label did not map cleanly to deep experience. Among roles with “Senior” in the title that also stated years of experience, 65% asked for five years or less, and the median was five years. For “Staff” roles, the median was 10 years.

Compensation data showed a shift toward more transparency. The report says 956 postings, about 15% of the dataset, included an explicit USD salary range. The median band was $150K to $210K, and the 25th to 75th percentile band was $120K to $250K.

The author also says 53% of postings were remote. That matters because the salary figures in the report are often shaped by globally distributed teams, not just local hiring markets.

Finally, the analysis looked at hiring language itself. It tracked 30 common fluff phrases and found that “self-starter” appeared in only 0.6% of postings, which the report says is far below what is often seen in corporate job descriptions. The full dataset, scrapers, and analysis code were published in the project’s data folder under an MIT license, according to the post.