StarterAfter Python Foundations
Tidy the folder you keep avoiding
Sort a downloads folder by type and date, rename files to a consistent pattern and write a log of everything that moved, so the change is reversible.
What you makeA script that sorts and renames, a log of every move and a dry-run mode that shows what would happen.
The question that tests itTwo files want the same name. Show what your program does instead of overwriting one of them.
StarterAfter Python Foundations
A report your spreadsheet builds for you
Read a CSV export, calculate the few numbers that matter and produce a formatted summary you could send to someone, without opening a spreadsheet at all.
What you makeA generator that reads the export, a formatted output file and a template you can change without touching the code.
The question that tests itThe export arrives with a missing column. Show whether your report fails loudly or produces a quietly wrong number.
StarterAfter Web Scraping & Automation
A watcher for a price you care about
Check a published price on a source you are permitted to use, keep the history and tell you when it moves beyond a threshold you set.
What you makeScheduled checks, a price history, a threshold alert and a record of every run including the ones that found nothing.
The question that tests itThe page changes and your selector stops matching. Show how you learn about that rather than assuming the price never moved.
StarterAfter Web Scraping & Automation
One digest instead of ten open tabs
Collect updates from a handful of permitted sources, drop the ones you have already seen and send yourself a single summary each morning.
What you makeMulti-source collection, a record of what you have seen, a scheduled summary and a delivery step.
The question that tests itOne source goes down. Show that the digest still arrives with the rest, and says what is missing.
StarterAfter Web Scraping & Automation
A tracker for the roles you want
Collect job postings from sources you are permitted to use, pull out the skills each one asks for, and compare them with the skills you can currently demonstrate.
What you makeScheduled collection, skill extraction, a gap list and a record of what changed week to week.
The question that tests itA site changes its layout and a run collects nothing. Show how you noticed and what you fixed.
StarterAfter Web Scraping & Automation and Advanced Python
Market data, and why a backtest flatters you
Collect published price history, calculate a few indicators and test a simple rule. Then add costs, slippage and a different time period, and watch the result change. The lesson is measurement, not stock picking.
What you makeData collection, indicators, a backtest, and the same test repeated with costs and a different period.
The question that tests itExplain why the first result looked better than the second. This is a teaching exercise in evaluation; it is not investment advice and recommends no strategy.
IntermediateAfter APIs & Integrations
A service that keeps what people send it
Accept a submission over an API, validate every field, store it and acknowledge it. Then handle the submission that arrives twice and the one designed to break you.
What you makeA validated endpoint, stored records, duplicate handling and tests for the inputs you did not expect.
The question that tests itSomeone submits the same form three times in a second. Show what you stored and why.
IntermediateAfter APIs & Integrations
A short link that tells you who clicked
Turn a long address into a short one, count the visits and show them on a small page. Simple enough to finish, deep enough to teach storage, redirects and counting honestly.
What you makeA redirect service, click records, a summary page and a decision about what you deliberately do not collect.
The question that tests itExplain what your counts include: retries, previews and your own testing all inflate a click number.
IntermediateAfter APIs & Integrations and Production Engineering
A scheduled pipeline that survives a bad day
Pull from an API on a schedule, process what arrives and store the result. Then make it behave when the source is slow, partly broken or returns yesterday's data.
What you makeScheduled runs, retries with backoff, a record of each run's outcome and a way to safely re-run a failed one.
The question that tests itA run half-finishes. Show that re-running it does not double the data.
IntermediateAfter APIs & Integrations and Production Engineering
A bot that tells you when something breaks
A messaging bot that reports a daily summary and raises an alert when a check fails, so you find out before someone else does.
What you makeScheduled checks, a messaging integration, alert rules and a quiet mode that prevents repeat noise.
The question that tests itMake a check fail repeatedly. Show that you are told once, not sixty times.
IntermediateAfter AWS for Developers
A watcher for your cloud spending
Read usage from your practice cloud account, show what each resource costs, and warn you before a forgotten instance turns into a bill.
What you makeUsage collection, a cost breakdown, a threshold alert and notes on shutting a lab down.
The question that tests itExplain which resource drove the largest change, and what you would turn off first.
AdvancedAfter APIs & Integrations and RAG Engineering
A document triage service
Sort incoming documents into categories, attach a confidence level and send anything uncertain to a human queue rather than guessing.
What you makeClassification service, a confidence threshold, a review queue and a record of decisions.
The question that tests itFeed it a document unlike anything it has seen. Show that it asks for review instead of choosing confidently.
AdvancedAfter APIs & Integrations and RAG Engineering
An assistant that answers from your database
Table-aware answering: the visitor asks a business question in plain words, the assistant writes the query, runs it and shows both the number and the query it used. Nothing is hidden.
What you makeQuestion-to-query step, a safe read-only connection, results shown with the query, and a set of test questions.
The question that tests itAsk an ambiguous question. Show how the assistant asks for clarification instead of guessing a number.
AdvancedAfter RAG Engineering
A knowledge graph of your documents
Pull the people, products and events out of a document set and store how they connect. Then answer questions that need two or three facts joined together, which a single passage cannot answer.
What you makeExtraction pipeline, a graph of entities and relationships, and questions answered by following connections.
The question that tests itAsk something the graph links wrongly. Show where the extraction went wrong and how you would catch it next time.
AdvancedAfter Agentic AI & MCP
A personal assistant for your working day
An assistant that reads a practice calendar and message export, drafts your day's summary and prepares replies. It waits for your approval before anything is sent.
What you makeScheduled summary, drafted replies, an approval step and a trace of every action considered.
The question that tests itRefuse a draft and change an instruction. Show that nothing left the system without your approval.
AdvancedAfter Web Scraping & Automation, RAG Engineering and Agentic AI & MCP
A research assistant that gathers, reads and reports
The whole roadmap in one project. It collects material from permitted sources, indexes it so answers can cite a passage, then works through a research question and writes up what it found, with the gaps named and a person approving before anything is published.
What you makeScheduled collection, a retrievable index with sources, a multi-step research workflow, cited findings and an approval step.
The question that tests itAsk it something the collected material cannot support. Show that it reports the gap instead of writing a confident paragraph anyway.