The Rise of the Cyber Defense Engineer
TL;DRWhy the most important job in cyber defense now belongs to an engineer
- Orion Cassetto
- 16 Nov 2026
- 5 minutes
Security teams were built around a scarce resource: skilled people with the hours to look at alerts. Tiers, shifts, escalation paths, and the tickets between them all exist to ration that expertise. Adversaries never had that limit. In November 2025, Anthropic reported that a state-sponsored group had an AI model run 80 to 90 percent of an espionage campaign against about thirty organizations, with people stepping in at a handful of decision points. Now defense doesn't have that limit either, because AI agents can do the investigative work. The analyst, detection engineer, and threat hunter merge into one role, the Cyber Defense Engineer. They write the detections, the skills that tell agents how to work each alert, and the business context behind them. Then they review the agents' reasoning and sign off on what changes.
Three jobs become one
The role draws on three jobs most security teams already staff. The SOC analyst brings the judgment of investigation: working an alert to a verdict, weighing it against how the business actually runs, and deciding what needs a person's attention. The detection engineer brings the discipline of turning that judgment into logic that runs, holds up, and stays tuned, and the threat hunter adds the habit of forming a hypothesis and testing it before any rule fires. In the traditional model, every detection passed from its author to someone else to work its alerts. The Cyber Defense Engineer owns that whole loop, from the first line of a detection to the verdicts it produces.
Go-to-market teams went through the same change a few years earlier. When AI and automation took over the research, sequencing, and personalization that filled a sales development rep's day, the people who designed that system became a profession of their own, the GTM engineer. The Cyber Defense Engineer is that profession's counterpart in security operations.
The engineer stays in charge
The agents take the first look at each new alert, including the lookups across multiple consoles that used to fill a shift, but they can't change what runs in production. A detection an agent writes is saved as a draft, a tuning change waits as a suggestion, and neither goes live until the engineer approves it. The engineer owns the decisions on risk and reviews each verdict, with its evidence and the steps the agents took, the way they would a colleague's case.
What a Cyber Defense Engineer does
The engineer starts by deciding what the organization needs to catch, writes the detection, and then adds the skill that tells agents which sources to check and what evidence settles each alert it raises.
The engineer also gives the agents the context a skill assumes: why a service account exists, which deployment looks unusual but was approved, which systems the business cannot lose. With that context, a single hypothesis from the engineer becomes a search across the whole estate and its full history, and what it confirms can become a standing detection.
One threat, start to finish
A single threat report is enough to run every step of that loop.
Hypothesis. A threat report describes attackers spreading through rogue copies of ScreenConnect, a remote-support tool IT teams use every day. The engineer hands the report to the agents, and the search across every endpoint comes back clean.
Detection. A clean hunt still leaves the door open, so the engineer turns on a detection for the behavior, and it fires on the help desk doing ordinary support.
Skill. The rule can't tell IT from an intruder. The engineer writes the skill that can: a session from the company's own support server is routine, and a connection to an unfamiliar host that launches scripts gets escalated.
Verdicts. A week later IT adds a new support server, and the agents flag its first session as suspicious. The engineer marks it benign with one line saying the server is IT's, and that context shapes how the agents read every session from it afterward.
The best candidates are already on the team
The detection engineer who keeps asking why a rule needs a person to finish it is usually on staff already, and so is the senior analyst whose case notes everyone else copies.
Start small. Give one of those people the noisiest detection in the environment, have them write the skill behind it, and record the facts the agents keep asking for. Measure what changes in verdict quality and hours returned, then move to the next noisiest one.
Built for the Cyber Defense Engineer
Vega's customers are the first of a new breed, the Cyber Defense Engineer, who builds and directs AI reasoning instead of chasing alerts. On Vega, swarms of AI agents detect, triage, and investigate threats across all of a company's data in place, without pipelines to maintain or logs to ingest. Engineers write their skills in Detection Skills, the open standard Vega published at detectionskills.io, and Vega Memory holds the context those skills assume. In Anthropic's published case study, Vega customers report reclaiming roughly 67% of their cyber defense engineering team's time.
"Retail runs on peak moments, and attackers know exactly when those are. My team cannot be the constraint on a Saturday in December. Detection Skills gives us leverage we can plan around: the expertise is written down, it runs on every alert, and it holds up when volume spikes. We are adopting it and sharing what we learn, because no security team should have to rebuild this work alone."
Shawn McGhee, CISO, Exemplar Luxury Group
Every security team that adopts AI agents will end up with someone deciding what they check and when they escalate. The teams that pull ahead will make that a named job and hand it to the engineers who already know their environment.
- #detection engineering
- #Detection Skills
- #AI agents
- #threat hunting
- #soc transformation
- #agentic SOC