1. LandOffer
  2. Blog
  3. All articles
  4. Application Security vs. Detection and Response Jobs: Choose by the Systems You Have Secured

Job search

Application Security vs. Detection and Response Jobs: Choose by the Systems You Have Secured

Four actual employer roles and a completed fictional preventive-control and detection handoff distinguish security responsibilities and candidate evidence.

10 min readLandOffer team

Your strongest security story might begin with an authorization decision that changed a product design, or with missing telemetry that made an incident difficult to reconstruct. Both require understanding how systems fail. When you apply for application security versus detection and response jobs, show the decisions you owned and the evidence those decisions produced.

Choose by the system you owned and the decisions you can explain. A scanner finding, a dashboard, or a security certification alone leaves important questions unanswered: who changed the design, who maintained the control, and who handled the production consequences?

A cutaway workshop connects a product design gate to an observation and response desk

Read the ownership verbs before the security title

Application security commonly involves reviewing designs, identifying trust boundaries, turning risks into engineering requirements, and following remediation into the product. Detection and response commonly involves deciding what events must be collected, writing and maintaining detection logic, investigating signals, and coordinating containment. Those are useful starting points for reading a posting; they do not create a universal organizational boundary.

Ask what the employer expects you to leave behind. A design review should change an implementation decision or document accepted risk. A detection should have a data dependency, an owner, a testing approach, and a response path. Explain what you built, changed, reviewed, or operated alongside the tools you used.

Use LandOffer's recent job search to collect relevant openings, then read each employer's complete requirements. Search terms help discover roles; the role body determines which experience belongs in your application.

What four current employer roles actually require

We checked these public employer postings and their rendered descriptions on October 20, 2026. This is a bounded comparison of four openings, not a measurement of the security job market. We did not interview the teams, apply, inspect their systems, or test their security controls. Listings and team arrangements can change.

Employer and role System or decision owned Evidence to prepare Boundary to keep
Smartsheet, Principal Security Engineer, 8157847 Design-stage product security reviews and threat-modeling practice Architecture reasoning that became enforceable requirements; influence across engineering Principal scope, with 10+ years requested; a single review is narrower evidence
Anthropic, Staff+ Application Security Engineer, 4502508008 Production security tooling and secure design for AI systems Durable engineering plus reasoning about trust boundaries Includes an operational rotation and incident response
Aircall, Senior Security Engineer, Detection & Response, 4396290009 Security data platform, detection lifecycle, and response Maintained pipelines, production code, investigations Building and operating infrastructure is central
Flexport, Senior Security Engineer, Detection & Response, 8164847 Detection quality, telemetry contracts, and incident closure Tested detection changes and defensible response decisions The team that writes detections also receives their pages

The Smartsheet posting ties architecture and data-flow reasoning to requirements before release. It places this principal role above isolated review execution: the engineer shapes the review service, technical standards, and team capability. Its location text permits Bellevue or US remote work where the company is a registered employer. Treat that geographic restriction separately from technical fit.

The Anthropic opening combines coding, product partnerships, and threat modeling with operating security systems. It explicitly includes bounty escalations, incident response, and launch consultation in a shared rotation. Its remote-friendly heading also says travel is required; the logistics section describes office attendance. Neither the AppSec label nor the remote-friendly wording removes those obligations.

At Aircall, the description estimates roughly two-thirds software and platform engineering, with the remainder covering investigation, response, and hunting. It names production Python, Terraform, ingestion infrastructure, retention and cost decisions, and an on-call rotation. The listed locations are its New York, San Francisco, and Seattle offices. This example makes a useful distinction between consuming alerts and owning the machinery behind them.

The Flexport US-remote role calls for versioned, reviewed detections and measured quality, including precision and time to detect. It also assigns prelaunch telemetry requirements and incident retrospectives. The posting describes a follow-the-sun pager arrangement; verify the actual coverage schedule rather than generalizing that arrangement to other employers.

Application security evidence should reach the implementation

An AppSec story becomes stronger when you can explain the original design, the trust assumption you challenged, the requirement that changed, and how the team checked the result. Connect the vulnerability class to your contribution so a reviewer can distinguish finding a problem from helping resolve it.

For example, distinguish identifying a missing authorization check from designing the ownership model, negotiating its rollout, writing regression cases, and confirming remediation. If another engineer implemented the change, say that. Collaboration is substantive work when you can explain the decision and your responsibility accurately.

Also distinguish a scan result from a confirmed product risk. Dependency presence, exploitability, exposure, and remediation priority are different questions. Describe what you actually established and what remained uncertain. Avoid taking employer code, internal architecture diagrams, customer identifiers, or vulnerability reports into a public portfolio. A clearly fictional reconstruction can preserve reasoning without claiming to reproduce a real assessment.

Detection work depends on data and response authority

A detection needs more than a plausible query. Explain which event source contains the necessary fields, what ingestion delays or gaps mean, how you evaluated expected behavior, and what a responder should do next. An alert with no reliable actor identity can be much harder to act on than its severity label suggests.

Keep quality claims scoped to their denominator. Reviewing 12 alerts does not establish recall across every incident. A quiet week does not prove a detection covers the intended behavior. If you measured precision on a labeled sample, give the sample and labeling method; if you only wrote test cases, describe those test cases.

Response ownership also involves authority. Who can disable a credential, interrupt a service, or approve restoration? Explain which owner approved a disruptive containment action, what evidence supported escalation, and what you were authorized to do. Production reliability experience transfers well here, but an availability incident is not automatically a security investigation.

Worked case: one export feature, two security deliverables

The following completed design exercise is fictional. It concerns an invented multi-tenant document service with a background export worker. No application was built or attacked, no target was probed, and no production logs or customer data were used. Its purpose is to show the different artifacts a candidate can discuss.

AppSec review result. The initial proposal checked tenant membership when a user requested an export, then let a worker fetch documents later. The review identified a time boundary: membership might be revoked before the worker executes. The completed requirement is to authorize the requesting principal against the requested tenant at execution time, bind the export job to that principal and tenant, and deny execution after revocation. The product engineer owns implementation; the reviewer owns the requirement and review closure.

The acceptance record contains three specified cases: a still-authorized requester receives only the intended tenant's documents; a revoked requester receives no export; a requester whose job references another tenant is denied. These are proposed acceptance expectations recorded in this fictional review; no tests were executed. This exercise does not establish that a real implementation has no authorization flaws.

Detection handoff result. The companion event contract records a pseudonymous principal identifier, tenant identifier, job identifier, authorization decision, decision reason, event time, and schema version. It excludes document content and download credentials. Access and retention require an explicit policy decision; omitting content does not make identifiers harmless.

The initial detection hypothesis is repeated denied export execution after membership revocation. The completed review below prevents a toy rule from being presented as a proven attack detector.

Fictional observation Interpretation Completed handling decision
One job emits five identical denial events because the worker retries Repeated transport activity can inflate counts Count unique job identifiers; retain retry count for diagnosis
Five distinct jobs for the same principal are denied after revocation Worth investigation under this exercise's hypothesis Open a human-reviewed investigation with the underlying event references
An event lacks the authorization reason The intended condition cannot be established Route to telemetry-quality review; do not silently treat it as evidence of abuse
An approved internal revocation exercise produces five distinct denied jobs Expected activity can resemble the signal Preserve the event trail and annotate the authorized exercise context

Five jobs is an illustrative threshold chosen for this exercise, not a recommended production setting. No precision, recall, latency, or attack-prevention result was measured. The design still needs a time window, late-arrival policy, storage limits, validated event provenance, and evaluation against representative benign activity before deployment.

The resulting response note is also complete: verify the membership change and job records, check authorized exercise context, escalate credible misuse to the incident lead, and request any disruptive containment through the approved owner. It does not automatically disable a user. AppSec supplies the preventive requirement; detection engineers supply observable evidence and an investigation path. Both teams must agree what a denial event actually means.

An export review notebook and an event ledger show separate decisions joined by a shared job identifier

Route your application using the work you can defend

Consider two fictional applicants. Elise maintained a tenant-permission library, reviewed export designs with product engineers, and helped close authorization regressions. She supported incident follow-up but has never owned a security ingestion platform. Her strongest route is an application or product security role whose level matches that scope. The principal Smartsheet opening additionally asks for sustained program and technical leadership; the example does not establish that she meets it.

Her evidence statement is specific: “Reviewed the export authorization design, specified execution-time membership checks, and partnered with the owning team on regression acceptance.” She should only add “verified” or a measured result when her actual records support it. She should not turn participation in one incident into incident-command experience.

Malik maintained identity-event ingestion, corrected duplicate-event counting, tuned reviewed detections, and coordinated investigation handoffs during his team's coverage hours. His strongest route is detection engineering or response engineering. Aircall's infrastructure-heavy description makes his production automation relevant, while his depth in Terraform and platform ownership still needs checking.

His statement might read: “Maintained identity-event ingestion and revised detection counting to distinguish retries from unique operations; documented investigation evidence and escalation ownership.” He should not claim an improved detection rate without a defined evaluation. If his experience was operational monitoring rather than security, he should describe that distinction and supply a separate, labeled security design exercise.

These routes are preparation decisions, not hiring predictions. Both applicants still need to compare level, location, work arrangements, and required systems experience with each current posting.

Questions that reveal the work behind the title

Ask an AppSec team how a design review enters the development process, who can require a change, and how closure is verified. Ask what proportion of the job involves tooling, manual review, vulnerability operations, and product consultation. Those answers determine whether your review experience matches the job's actual ownership.

Ask a detection team who owns ingestion failures, detection releases, pages, and containment approvals. Request the coverage hours and escalation model, including handoffs. A role with extensive pipeline responsibility needs engineering evidence beyond familiarity with a SIEM interface.

For either team, ask how AI-assisted findings or response suggestions are reviewed and who accepts the risk of automation. The useful answer names a human decision, an evaluation process, and an operational owner. Use the answers to compare the team’s actual responsibilities with work you can substantiate.

Prepare a small evidence packet before applying

Choose one system you know well, write a short account of its security boundary, and identify your actual contribution. Add one sanitized artifact and one limitation you can explain without revealing private information. A design decision, a reviewed detection change, or an incident handoff is more useful than a portfolio full of unexplained screenshots.

Compare that packet with one opening’s deliverables, mark the gaps, and keep only claims you can defend in an interview. Use LandOffer's recent openings to continue the search with a clearer target: the security work you have performed, the systems you can discuss, and the responsibilities you are ready to own.

Sources and Further Reading