KeMeT Tech
← All practice areas
05 / SIEM & SOAR

Detection engineering

Detection rules and SOAR playbooks that fire on real attacks, not on noise.

What we ship

Detection rules tied to MITRE ATT&CK techniques with coverage gaps documented. SOAR playbooks that triage, enrich, and only escalate when human attention is warranted. Custom connectors when the vendor option does not exist.

On Microsoft Sentinel: analytics rules with KQL transforms at ingestion to cut noise, watchlists for asset context, Logic Apps for SOAR, MMA-to-AMA migration done right (see our field notes on this).

Coverage versus noise

More rules is not better. We start with what you cannot afford to miss and work outward, retiring rules that fire on benign activity. Every engagement includes a rule retirement list: detection rules currently in production that are pure noise.

Multi-SIEM experience

We have shipped on Sentinel, Splunk, Chronicle, Elastic, and QRadar. When the question is 'which SIEM,' the answer depends on your existing data sources, license posture, and team. We have written up the comparisons.

Free starter

Six production KQL rules, MIT licensed.

The detection-engineering engagement starts with the same six rules we open-sourced last quarter. Pull the pack, run it against your tenant, see the noise pattern before the discovery call.

What you receive

Deliverables, by line item.

  • Detection rule pack mapped to MITRE ATT&CK
  • SOAR playbooks with triage logic and human-loop checkpoints
  • Custom connectors where vendor options do not fit
  • Rule retirement plan with noise metrics
  • On-call runbook for the top ten incident types
How the engagement runs

Four steps, no mystery.

  1. 01
    Discovery call

    Thirty minutes with the senior engineer who does the work. Your environment, your constraint, your deadline.

  2. 02
    Written scope in 48h

    Deliverables, timeline, and one fixed price, in writing. Sign it through your normal procurement flow.

  3. 03
    Build in your tenant

    Work lands in your repos and your cloud accounts from week one. A demo of working output every week.

  4. 04
    Handoff, documented

    Runbook, architecture decisions, and a working session with your team. You own everything we built.