
Here is a question I get from people building a name for themselves in tech: “What can I put on my public profile that proves I think like an engineer, and not just a certificate collector?” One answer I like more and more is a small, well-kept repository of detection rules, the kind of thing described by the term DetectionOps. Before you roll your eyes at yet another “Ops” word, give me a few minutes. I will explain what it means in plain language, why it suits a personal portfolio, and how to avoid the traps that make this kind of project look hollow.
The plain-language version
Imagine your phone’s spam filter, but for servers. Someone has to decide what counts as suspicious, write that decision down in a form a machine can follow, and then keep checking that it still behaves. Each of those decisions is a detection rule. DetectionOps means looking after the rules with the same routines a software team uses: a written reason for each rule, the rule stored as code, tests, a second pair of eyes, a controlled release, and an honest look at how it performs afterwards. The Ops Atlas page about DetectionOps lays this out and adds a candid footnote that the name is informal and not an official standard. The respected material on the topic uses other labels.
That footnote is helpful for personal branding. Using the established terms in your write-ups shows you did your homework, and mentioning the newer label shows you are paying attention.
Why it makes a good public project
A portfolio project should be small enough to finish, visible enough to be judged and rich enough to talk about. Rule repositories tick all three boxes. Every rule is a few lines. Every change shows up in the history. And each one invites a story: what behaviour I targeted, how I tested it, what went wrong, how I fixed it.
Contrast that with a screenshot of a certificate. A certificate says you passed an exam, while a repository says how you work. If you have read our earlier tips on typical Docker slip-ups that beginners can sidestep, you will recognise the theme: employers trust evidence of careful habits more than lists of tool names.
What goes into the repository
Keep it modest, and in this order.
- A README with the scope. State the boundaries up front. The Atlas is tidy about them: rules, requirements, tests, coverage evidence and tuning records are in; running an alert queue, hunting for threats, handling incidents and collecting logs are out.
- One short requirement per rule. A paragraph naming the behaviour, the log data needed and the harmless look-alikes.
- The rule itself. Sigma is a good open format to practise, which asks for a name, the log type searched and a matching condition, but the entry says it is not compulsory, so a native query kept as code is acceptable.
- Two tests. One sample that should fire the rule and one that should stay silent, written from the behaviour, not from the rule.
- A pilot note. What happened when the rule ran quietly on sample data for a few days.
- A numbers page. Precision for each rule, plus anything you tuned.
The numbers, explained simply
Precision is just the proportion of alerts that were worth having. If a rule raised forty alerts and only six were real, as in the Atlas’s worked example, its precision is fifteen per cent, and each real catch cost an analyst about forty minutes. Showing that you can calculate and improve a figure like that lands better with a hiring manager than a profile line announcing enthusiasm for security.
The same example makes a second point that I love for personal projects: a tag is not proof. Labelling your rules with technique codes from the public ATT&CK catalogue is fine, but claiming broad coverage because of the labels is not. In the entry’s illustration, labels suggested complete coverage, but staged attacks showed sixty per cent of the prioritised cases were actually caught. If your portfolio distinguishes the two, you stand out immediately.
Traps that make a portfolio look weak
- Copying a big public rule pack. The shared Sigma library has more than three thousand rules; the Atlas calls them starting points. Adapting five and explaining why beats importing a thousand.
- Tests that echo the rule. They pass by construction and prove nothing.
- Silent exclusions. Every filter reduces noise by hiding something. Note what is hidden and when the exception ends.
- Floating tool versions. The entry mentions a case in October 2026 where a new major release of one library clashed with tools still needing the older line. Pin your versions, as the container habits in our guide to container practices for development and production also advise.
- Overclaiming. Do not describe your lab as a production programme.
How to tell the story publicly
Write short posts, one per rule. Open with the behaviour in everyday words, show the before-and-after precision, and admit what is still a gap. Admitting gaps is not weakness; the Atlas’s own method turns accepted blind spots into named tests that someone owns and that lapse on a set date, so being open about them mirrors professional practice. When you share a link, share the repository history, not just the finished state, because the history shows how you think.
Be careful with claims about the wider industry. The entry’s survey figures, for example 45 per cent of 264 respondents reliably measuring detection effectiveness in a 2025 study, are fine to cite with the source and sample size, but should not be stretched into sweeping statements. Likewise, the only case the Atlas documents is a government security team describing its own programme, which nobody outside has independently assessed.
A ninety-day plan for the busy person
Month one: set up the repository, add a linter, and write the README. Month two: write three rules with requirements and two tests each. Month three: run a quiet pilot, compute precision, publish a short write-up and ask a colleague to review a rule through a pull request. The entry suggests a similar month-by-month rhythm for teams; this version is trimmed to fit evenings and weekends.
At the end you will own a small, honest body of work. It will not make you a security expert overnight, but it will show how you handle rules, evidence and change, and that is precisely the story an engineering reputation is built on.







