Search Engineer Trainee
Use each lab as an experiment. Follow the mission steps, then observe the system response.
- Overview: watch one query move through all four layers.
- Crawl: ensure important public docs enter the repository but private/test docs do not.
- Index: build posting lists and run an intersection query.
- Rank: adjust signal weights and see which doc becomes #1.
Mission: click "Animate" and carefully read each box as it lights up.
1. Crawl
2. Analyse
3. Inverted index & rank
4. Query & infra
1. Press the button once and track the order of boxes.
2. Say out loud: "crawl → analyse → index & rank → query & infra".
3. Ask: where do my documents live? Where does my query travel?
Mission: allow important docs, block login/test docs, then run the crawl.
1. Keep doc101 and doc102 crawlable.
2. Block doc103 (login) and doc104 (robots/test).
3. Decide what to do with doc105 (low‑value parameter page).
4. Click "Run crawl & analyse" and check the log.
Mission: make sure the key documents appear in postings for "search" and "index".
1. Click terms in each document to send doc IDs into postings.
2. Use the term selector to inspect postings.
3. When ready, press "Run sample query" to intersect postings for "search" and "index".
Mission: get the most balanced technical article to rank #1.
1. Start with all signals on and note the top document.
2. Turn off PageRank and see what changes.
3. Turn off freshness and compare again.
4. Discuss which combination seems best for this query.
Mission: produce a good candidate set while keeping latency under ~80 ms.
1. Choose which query features should be on for "search engine design".
2. Adjust server sliders for India, Europe and US.
3. Run the query and check candidate docs and latency.
4. Try to reach a configuration with both good docs and low latency.