Upgrade & Secure Your Future with DevOps, SRE, DevSecOps, MLOps!
We spend hours scrolling social media and waste money on things we forget, but won’t spend 30 minutes a day earning certifications that can change our lives.
Master in DevOps, SRE, DevSecOps & MLOps by DevOps School!
Learn from Guru Rajesh Kumar and double your salary in just one year.
Anyone who works with machine learning and operations data knows the odd feeling of staring at a notebook at four in the afternoon and realising that nothing on the screen makes sense. The model did not get harder. You simply ran out of fuel. A friendly fix is to match tasks to your energy levels: give the thinking-heavy jobs to the hours when your head is clear, and save the mechanical ones for the hours when it is not. Here are nine ways to do that in an AIOps routine, each small enough to try this week.
1. Notice your own rhythm before changing anything
Do not copy somebody else’s schedule. Spend a few days jotting a quick one-to-five score for how alert you feel at a handful of points in the day. Most people find one clear peak and one clear dip. Yours might be early, late, or oddly split around lunch. Everything else on this list depends on knowing it.
2. Give feature design your sharpest hour
Choosing which signals feed a detector, and how they are windowed or normalised, is a lot of juggling. You hold the shape of the data in your head while you weigh trade-offs. That is peak-hour work. If you tackle it while drained, you will pick the first idea that looks acceptable and regret it during the next incident review.
3. Read incident timelines when you are fresh
Going back over what a model saw, what it missed and what a human noticed first takes patient pattern-spotting. Doing it on a good morning means you catch the small oddities, like a metric that moved a minute before the alert did. Doing it at the end of a long day means you skim and agree with the first summary you find.
4. Let the slump handle label clean-up
Backlogs of alerts that need a quick tag are ideal for the low-energy window. The decisions are small and similar, and you do not need to invent anything. Put on some music, work through a pile, and call it done. You will be surprised how much of this you can clear in a sluggish hour that would otherwise be wasted.
5. Bump dependencies when your attention is flat
Updating libraries, regenerating lockfiles and rerunning the test suite are mostly waiting and clicking. They still need care, but not creativity. Scheduling them in the dip protects the morning for work that actually benefits from a clear mind, and the dip gets a purpose instead of being something you push through.
6. Match meetings to the middle of your curve
Conversations about priorities and hand-offs need you to listen well and speak clearly, which is neither your sharpest nor your dullest mode. If you can, nudge syncs to the shoulder of your day. Where you cannot move them, at least avoid stacking three of them across your peak.
7. Write documentation in two passes
Docs are a good example of a task with two energy levels inside it. Outlining what a runbook should say, and in what order, is demanding. Fixing broken links, tidying headings and pasting screenshots is not. Do the outline in a good hour, and leave the cosmetic pass for later. For a tour of how a team can approach this, a post on getting AIOps adopted by an IT team shows the kind of planning that benefits from your best attention.
8. Keep a short list of dip-friendly jobs
The biggest trap in the slump is not having anything to do that suits it, so you drift into email or doomscrolling. Keep a visible list of low-effort, low-risk tasks: renaming dashboards, archiving stale notebooks, checking that a scheduled job still ran. When your energy drops, you pick from the list instead of deciding what to do, which costs energy of its own.
9. Protect one demanding task per peak
It is tempting to fill your best hours with three ambitious things. Pick one. Name it the night before, close chat while you work, and finish it or reach a clear stopping point. A single well-served demanding task beats three half-served ones, and it gives you a simple measure of whether your day went well.
Where energy matching stops working
Real life interrupts. On-call weeks, production surprises and other people’s deadlines will ignore your curve, and that is fine. The point is not to build a perfect timetable. It is to have a default that you return to whenever you get a choice. Also remember that energy is affected by sleep, food, movement and stress, so your pattern can shift between weeks. Re-check your notes now and then rather than assuming the old answer still holds.
A second caution: some tasks mix both kinds of energy, and trying to slice them too finely creates its own overhead. If you spend more time classifying a job than doing it, just do it. The sorting is meant to be rough. Think of three buckets, such as sharp, steady and slow, and place work into them by feel.
A sample day for an AIOps engineer
To make this concrete, picture a person who feels most alert in the late morning and flattest after lunch. They begin with a short, low-key warm-up: scanning overnight alerts and making notes without trying to resolve anything. Then, in the peak, they spend a solid stretch redesigning how a seasonal pattern is handled in a forecasting job. After lunch they work through label clean-up and a dependency bump. Late in the afternoon, their energy ticks up a little, so they use it to write a clear hand-over note for the evening shift and to pick tomorrow’s single demanding task.
Nothing in that day is dramatic. The engineer is not working longer, only arranging the same tasks in a more sensible order. At the end, the demanding work is finished and the chores are done, and the person has some energy left over for life outside the job.
A note on civic life and your calendar
Energy matching is not only a work habit. It applies to the small admin tasks of ordinary life too, such as writing to a local representative about an issue you care about. Drafting that message well takes a clear head. One option some people look at is JanBoard, an independent site that hosts appeals which anyone can read, addressed to elected representatives and public bodies; by its own description it is not run by the government. Whatever route you pick, doing the drafting in a good hour and the routine follow-up in a tired one is the same principle.
Try it for one week
Pick three items from the list above and run them for five working days. Keep your quick one-to-five scores going so you can see whether the peak and dip are where you thought. At the end of the week, ask two questions: did the demanding tasks feel less like wading through mud, and did the chores feel lighter when they met a flat afternoon? If the answer to both is even a mild yes, you have found something worth keeping. If not, tweak the placement and try again. For more everyday approaches to getting started, a gentle read on tackling live technical faults for newcomers sits well alongside with the habits above.
The goal is modest and kind to yourself: get more of the important work done with less strain. Your energy will always rise and fall. A little planning lets you work with it instead of against it.