reference · sources

Sources: collect data yourself

Ready CSVs are convenient, but the real skill is pulling fresh, messy data yourself. Start with the ones that need no key.

APIs — collect data yourself

Ready-made CSVs are comfy, but the real skill is pulling fresh, still-messy data yourself — just like on the job. Start with the ones that need no key, and don't chase the 'correct' API — grab the one whose topic you actually care about.

Public APIs (каталог)

GitHub list · thousands of APIs

A giant catalog of free APIs by category — weather, finance, sport, games, animals — each tagged with auth and rate limits. Your first stop for 'is there even an API for X?'. There's also a hand-picked set of 10 APIs aimed at data engineering, with a ready pipeline example.

Tip: Open the catalog, jump to a category that grabs you, and pick the first no-key API. Interest beats 'correctness' — start with what excites you.

PokeAPI

no key · REST

Everything Pokémon: stats, types, evolutions, and the links between them. The perfect first-ever API — no signup, clean structure, plenty to cut your teeth on with JSON and requests.

Tip: Try it right now: curl https://pokeapi.co/api/v2/pokemon/pikachu
Questions
  • Which Pokémon type is strongest on average by stats?
  • Who has the longest evolution chain — and how do they branch?

REST Countries

no key · no limits

A reference on every country: population, area, currencies, languages, borders. Pairs beautifully with other data — enrich your dataset by country and build maps.

Questions
  • Which regions pack the most official languages per country?
  • Is there a link between a country's area and how many neighbours it has?

GitHub REST API

key optional · generous limits

Repo activity: commits, stars, issues, releases. Dig into open-source popularity and tech trends on the repos you actually love.

Questions
  • How did your favourite project grow in stars week by week?
  • Which days and hours see the most merged PRs?

Spotify Web API

key required · free

Tracks, artists, playlists, and audio features — danceability, energy, tempo, mood. Pull your own playlists and take your own taste apart piece by piece.

Questions
  • How does your favourite genre differ in audio features from the rest?
  • Does the mood of your playlists shift by time of day?

OpenWeather

key · 1 req/sec free

Current weather and forecasts worldwide. A classic when you want to cross weather with something else — flight delays, sales, how active people are.

Questions
  • Does weather move something you already collected — sales, trips, activity?

GDELT

world events · huge

A database of news events from all over the world, with sentiment scoring and live translation. The scale is enormous — for ambitious projects on global trends and tone.

Questions
  • How did the emotional tone of news on a topic you care about shift over a year?

Streaming — live data

Advanced stuff — save it for later, but these are the projects that make you stand out. Data flowing right now, plus event generators so you can build pipelines like a real product's.

Awesome Real-Time Datasets

GitHub list · dozens of feeds

A curated list of live feeds: transit, seismic, aviation, crypto, IoT sensors, threat intelligence. Plus synthetic generators that push events straight into Kafka. The easiest place to find a live source for whatever you're into.

Tip: Nothing clicks here? Check the sibling hub: github.com/ColinEberhardt/awesome-public-streaming-datasets

Wikimedia EventStreams

SSE · no key

A live stream of Wikipedia edits. The gentlest 'first stream' there is: connect and instantly watch events fly by. Perfect for grasping how real-time actually works.

Questions
  • How many edits per minute, and in which language editions?
  • Who's busier right now — bots or humans?

USGS Earthquakes

GeoJSON · updates live

A live feed of earthquakes worldwide: time, place, magnitude, depth. Constantly refreshed — great for real-time maps and simple alerts.

Questions
  • Where was the strongest shaking in the last 24 hours?
  • Is magnitude linked to how deep the quake is?

EventSim fakes a music service's event stream (think Spotify), while Lenses Datagen publishes ready datasets into Kafka — ship positions, taxi rides, IoT. You control the volume, so you can build a product pipeline without a real product.

Questions
  • What are your 'product's' DAU/WAU, and how are they trending?
  • Where do users drop off in the signup → activation → payment funnel?