How I Tracked the Tech Job Market as a Working Student (And What I Found)

The Setup: Shifting from Endless Scrolling to Automation

As a working student, I wanted to understand what the tech/data job market actually looks like right now. What skills do companies actually want? What tools are they using? Where are the opportunities?

So I did what any data person would do — I built a pipeline to find out.

I tried a few different job APIs along the way. Some expensive, some legally grey for a public post 😉. Therefore for this demo I'll show you HackerNews — feel free to swap it for whatever works for you, because with dltHub + AI it's so fast to build and then deploy that it honestly doesn't matter which one you start with 🙂

What is HackerNews “Who is Hiring?”

Every month YCombinator posts a "Who is Hiring?" thread on HackerNews. Real companies, one post each, no recruiters allowed. This month had 498 posts.

The data is completely free and open via the official HackerNews API and goes back to 2011, so you can track trends over time.

I pulled 3 months of threads (Apr–Jun 2026) — 987 job posts total — and used Claude to extract tools, roles, locations and remote status from each one. My pipeline is deployed on dltHub so it keeps running automatically every month and the dashboard is live there too, updating as new data comes in.

How I built it

I opened dlt AI Workbench, described what I wanted, and let Claude figure out the rest.

I didn't even have to think of the most interesting question. Claude suggested it: "Is AI bleeding into data engineering job requirements?" That became one of the core charts.

When I said I wanted it to run every month without re-processing old data, Claude designed the whole incremental loading pattern using dlt. First run of the month costs ~$0.40. Every re-run costs $0.

Claude built both a current snapshot and a trending view from day one so the dashboard automatically becomes more valuable every month.

Claude handled the login, the deployment, everything. One conversation.

You can also deploy directly with one terminal command: dlthub deploy

What the data shows

I'm not going to show you every chart but you can see the live dashboard — Python is #1, AWS dominates, you already know. Here's what actually surprised me:

Claude is now mentioned in 5.3% of all job posts — up from 3.6% in April and overall has the highest share the whole time anyways. Copilot is reducing and Cursor is volatile.

Some companies aren't just using tools — they're building the infrastructure layer itself. File systems, database engines, agentic development environments. This is the layer that doesn’t show up in top tools charts.

Only 3 months of data so far. Come back in 6 months for more updates 😊

Claude Haiku extracted these descriptions automatically from 987 job posts as part of the pipeline. Hover over any row to read the full description in the end of the dashboard.

Run it yourself

Fork the repo, open it in dltHub and ask Claude to change whatever you want — different roles, different API, different countries.

You can then deploy to your dltHub workspace by asking the agent - your dashboard stays live and updates itself.

Try dltHub free — 14 days, $30 in credits, no card required.