Python

Python for security and operations jobs: real Python 3.14 in the browser, marked on what your functions return and the state they leave.

About 59 hours · 29 modules · 353 labs · certificate on completion · Scripting and automation

For: analyst, engineer, administrator

Skills: python, automation, data-handling, api-integration

What the course covers

  1. 0. How Python runs here — Write a function, hand back its answer, read the traceback when it breaks, and know what the sandbox will not do. (9 labs)
  2. 1. Values, names and types — Read one morning's asset export the way Python reads it: text that looks like a number, zeros that look like nothing, and halves that round the other way. (11 labs)
  3. 2. Decisions, loops and functions — Write the checks an administrator runs every Monday as functions that decide in the right order, stop when they should, and cannot be called the wrong way. (11 labs)
  4. 3. Lists, dicts and sets over security events — Filter, count, group and rank a day of sign-ins, and hand back the data you were given exactly as it was. (13 labs)
  5. 4. Text and regular expressions — Turn auth.log, syslog and access-log lines into records, with patterns that match the field you mean and nothing longer. (13 labs)
  6. 5. Files and formats — Read CSV, JSON, JSON Lines and YAML the way they really arrive, and write files a second run does not spoil. (13 labs)
  7. 6. Errors and robust scripts — Skip the bad record, stop on the broken format, clean up on every path, and tell the people and the scheduler what happened. (12 labs)
  8. 7. Structure and tests — Import the team's modules instead of copying them, return records with names, and write the test that proves a bug and keeps it fixed. (12 labs)
  9. 8. Time — Put ISO 8601, epoch seconds and milliseconds, FILETIME and local-time tickets on one clock, through the night the clocks went back. (13 labs)
  10. 9. Counting and grouping at scale — Count, group and window the Detection course's labelled corpora with collections and itertools, and read a log too big to hold one line at a time. (12 labs)
  11. 10. pandas for analysts — Read scanner and SIEM exports as they really arrive, then filter, group, join and bucket them without losing a row or inventing one. (12 labs)
  12. 11. Calling APIs properly — Call a REST API the way an integration that runs unattended must: the key from the vault, every page, the server's own pace, and every ticket once. (13 labs)
  13. 12. Normalising security telemetry — Map Windows Security, Sysmon, Zeek, CloudTrail and Entra ID sign-ins to one schema, keep every raw record beside its normalised form, and flag what you cannot read instead of dropping or guessing it. (12 labs)
  14. 13. Detection logic in code — Write the threshold, the window and the sequence as functions, and prove each one on labelled events: every attacker caught, no benign twin fired. (14 labs)
  15. 14. Detection as code — Rules as YAML files, a loader that refuses what YAML quietly changed, a test per rule over labelled samples, a CI job whose exit status blocks the merge, and a review that reads a change by its effect. (12 labs)
  16. 15. Enrichment and hunting — Query the SIEM to its last page, ask a rate-limited reputation service once per indicator, and run a hunt in pandas that does not count anything twice. (12 labs)
  17. 16. Triage automation for the SOC — Group, de-duplicate and auto-close a day's alerts under guardrails, write a note an auditor can follow, and never close the incident that looks like the noise. (12 labs)
  18. 17. LLM-assisted triage, behind a validation gate — Put a simulated model's alert summaries through a gate that parses them, checks their shape and every claim against the alert's evidence, screens the evidence for planted instructions, and routes every failure to a human with the reason, closing nothing on the model's word. (11 labs)
  19. 18. SOAR: writing an integration — Write the connector a playbook calls: a class handed its client, vault and sleep, every page pulled, one written retry policy, errors with fields a caller can read, and a log the on-call can act on with no key in it. (12 labs)
  20. 19. SOAR: response actions, safely — Isolate, disable, revoke and delete as a playbook does: plan first, leave the allowlisted and the twin alone, ask where a person decides, and do nothing twice. (12 labs)
  21. 20. SOAR: case and ticket automation — Turn alerts into service desk tickets with keys that find the last run's work, so two runs leave exactly what one would, with SLA timers judged in UTC and closure notes an auditor can read. (12 labs)
  22. 21. Cloud: boto3 on the simulated AWS — Call AWS from Python the way a job that runs unattended must: the session it is handed, a client in every Region, every page, a waiter rather than a sleep, errors by their code, and credentials found the way botocore finds them. (13 labs)
  23. 22. Cloud: posture audits in code — Find what an auditor would find in an AWS account -- public buckets, old keys, missing MFA, open admin ports, logging that is off -- and not what a careless script would find instead. (12 labs)
  24. 23. Cloud: remediation as a Lambda — Write the function that closes a public bucket, an open port and a stray key the moment the event arrives, and leaves alone what is public by design. (12 labs)
  25. 24. Vulnerability management: scanner exports in code — Read a .nessus file and a Qualys-style CSV, make two scanners one list, and rank it by KEV, EPSS and CVSS with due dates in UTC. (12 labs)
  26. 25. Vulnerability management: ticket sync and KPIs — Port VM Lab 18's sync to Python and run it twice, close only what a rescan proves fixed, and compute MTTR and SLA compliance with pandas. (12 labs)
  27. 26. AppSec tooling — Read two scanners' SARIF as the format means it, check dependencies against an advisory feed with versions compared properly, and turn findings into tickets exactly once. (13 labs)
  28. 27. Administration automation — Parse the output of df, ps, systemctl and Get-Service, read INI and JSON, inventory several hosts, and turn an HR sheet into account changes with a dry run and a change log. (14 labs)
  29. 28. Shipping it, and the capstone — Package a detection job, give it a command and an exit status, configure it in layers, keep its secret out of the log, and make its tests and review rules the gate it ships through. (12 labs)

Free with an account.

Start for free