Open Positions

Offensive security is judged by evidence, so we hire for the habits that produce it: a stated starting position, a test that could fail, a conclusion another engineer can rebuild. Each opening below describes the work as delivered here, including the parts that are slow, procedural and unglamorous. Read the expectations closely. They are written to be assessed.

Hong Kong · Applications are administered via our careers portal. Shortlisted candidates meet the consultant who would supervise the work.

01

Junior Red Teamer / Penetration Tester

Full-time · Hong Kong

Begin an application

You will work inside authorised client environments alongside a senior consultant, on engagements that have already been scoped, approved and scheduled. The role is built around one capability: turning a loose observation into a result another engineer can reproduce from a written starting position. Ingenuity earns its place only when the evidence survives review, so expect to record what you tried, what failed, and what the failure ruled out.

What the role covers

These are standing duties rather than a project list. Work is assigned against them, reviewed against them, and used as the basis for the technical conversation at interview.

  1. 01Assess web applications, APIs, networks or cloud estates within the agreed scope. Record the identity used, the privilege held, the software version and the configuration that made each result reproducible.
  2. 02Form a hypothesis identifying the control obligation that should prevent an action, and design the minimal test capable of distinguishing genuine failure from false positive. Retain negative results: they define the authority the system still enforces.
  3. 03Study adversary techniques in the controlled laboratory before proposing them on client infrastructure. State the environment assumptions a technique depends on and the condition that would invalidate it.
  4. 04Maintain assessment utilities under version control so a reviewer can read exactly what executed. Pin dependencies, record invocation parameters and justify any deviation from a tool’s default behaviour.
  5. 05Write findings that name the failed control, the demonstrated consequence and the evidence holding the two together. Separate what was observed from what was inferred, including where an inference is weak.
  6. 06Retest agreed corrections under identical recorded conditions, and state whether the remediation eliminates the path, constrains it, or merely interrupts the technique originally exercised.
  7. 07Escalate unexpected system instability, a possible live compromise or any instruction beyond your authority before continuing. Authorisation is checked before an action, not explained afterwards.
  8. 08Take part in internal review, where another consultant rebuilds a finding from your notes alone. Their difficulty in doing so is treated as a defect in the record.

What we look for

We look for evidence of each capability rather than a claim to it. Where a requirement cannot be met, say so in the application: a stated gap is easier to assess than an implied one.

  • OSCE3, comprising OSEP, OSWE and OSED. BSCP, CRTM, OSIR or CREST accreditation strengthens an application but does not replace demonstrated work.
  • Working knowledge of Windows, Linux and Active Directory sufficient to explain how a limited account could acquire broader authority, and which telemetry would record that transition.
  • Fluency with the tools in common use, such as Burp Suite Professional, BloodHound, Nmap or Impacket, together with an understanding of where each one misleads its operator.
  • Ability to read and modify utilities in at least one of Python, PowerShell, C#, Go or Bash, and to explain the behaviour of code you did not write.
  • A clear distinction between laboratory research and action on a production system. You understand why request throttling, approval gates and accurate timekeeping change the risk of the same technique.
  • Written English precise enough to describe a technical condition to an engineer and to restate its business consequence without exceeding the available evidence.
  • A sanitised research sample stating the question, tested version, method and conclusion. Explain the root cause and identify precisely what the experiment did not establish.
  • Ability to distinguish an unsafe configuration, a demonstrated attack path and historical evidence of intrusion. Each claim carries a different burden of proof.

02

UI/UX Designer

Full-time · Hong Kong

Begin an application

You will design interfaces for people who interpret technical evidence and act on it under time pressure: an analyst reviewing an alert, an engineer reading a finding, a client approving a retest. The design problem is rarely visual. It is deciding what a screen should assert, what it should qualify, and when it should stop a user from proceeding. A good interface here makes uncertainty visible rather than resolving it on the user’s behalf.

What the role covers

These are standing duties rather than a project list. Work is assigned against them, reviewed against them, and used as the basis for the technical conversation at interview.

  1. 01Interview the people who use a workflow before defining it. Establish the decision the interface supports, who holds authority for that decision and the cost of resolving it wrongly.
  2. 02Map the states a task can occupy: permitted, restricted, incomplete, failed, irreversible. Design each one explicitly. A screen that only covers the successful path is unfinished.
  3. 03Produce wireframes, prototypes and implementation specifications precise enough for an engineer to build without guessing intent. Name interaction rules, not only visual properties.
  4. 04Run usability studies with technical and business readers. Record what participants did and where they hesitated before proposing a change, so the change can be justified later.
  5. 05Maintain components with documented behaviour. Review the implemented interface against the specification rather than reviewing a mock-up on its own.
  6. 06Design for keyboard operation, assistive technology and narrow viewports as baseline conditions. Justify trade-offs by observed user behaviour rather than aesthetic preference.
  7. 07Keep design decisions tied to security semantics: permission states, confidence levels, audit trails and irreversible actions must read as what they are, never as decoration.

What we look for

We look for evidence of each capability rather than a claim to it. Where a requirement cannot be met, say so in the application: a stated gap is easier to assess than an implied one.

  • Three or more years of professional product design, ideally on enterprise, operational or analytical software where density and correctness matter.
  • A portfolio that explains the problem, the research behind it, the iterations and the shipped result. Show a decision you reversed and what changed your mind.
  • Fluency with Figma or a comparable tool for maintainable prototypes, components and specifications.
  • Understanding of how a design system is implemented and maintained with engineers, and working knowledge of the WCAG principles that apply to dense interfaces.
  • A case study involving uncertain or conflicting evidence: show how the interface prevented a plausible misreading instead of merely presenting the data.
  • Willingness to learn security fundamentals: authorisation models, evidence standards, and the difference between a signal and an incident. Confidential client context must be handled with care.

03

Web Developer

Full-time · Hong Kong

Begin an application

You will build and maintain the applications that carry our work: client portals, reporting tools, assessment utilities and the internal services behind them. The distinguishing requirement is not framework familiarity. It is the ability to recognise a security boundary, place authorisation on the server side of it, and write the test that fails when the boundary moves.

What the role covers

These are standing duties rather than a project list. Work is assigned against them, reviewed against them, and used as the basis for the technical conversation at interview.

  1. 01Deliver React features across implementation, review and deployment, including loading, empty, error and access-denied states encountered in operation.
  2. 02Design services with explicit input validation, bounded queries and predictable error responses. Record the compatibility assumptions at each API boundary.
  3. 03Enforce authorisation on the server with least privilege, and treat proof of identity and permission as separate checks. Neither may be inferred from a client-supplied value.
  4. 04Measure performance against representative workloads and explain the query plan, deployment constraint or dependency responsible for any bottleneck.
  5. 05Write tests that detect meaningful regressions, including tests that demonstrate a security property such as tenant isolation, replay rejection or transaction atomicity.
  6. 06Participate in code review as a reviewer and as an author. Record significant assumptions so the next engineer can change the feature without breaking it silently.
  7. 07Reproduce consultant-reported defects from the recorded state and demonstrate that the correction restores the expected behaviour.

What we look for

We look for evidence of each capability rather than a claim to it. Where a requirement cannot be met, say so in the application: a stated gap is easier to assess than an implied one.

  • Strong React skills in JavaScript or TypeScript, with practical experience of component design, state management and data fetching.
  • Node.js services or integrations that run reliably outside a local development environment, including how they fail.
  • Understanding of API design, data modelling and database access rules, sufficient to predict how a query behaves as the dataset grows.
  • Secure development habits applied to untrusted input, dependencies, secrets and logging. Cloud or container deployment experience helps.
  • Ability to articulate a specific security property and define the test that would fail if that property ceased to hold.
  • Judgement about production records: how to support an investigation without capturing credentials or unnecessary personal data.
  • Written communication clear enough for a reviewer to follow a design decision without a meeting.

04

Marketing Promoter

Full-time / Part-time · Hong Kong

Begin an application

You will explain offensive security work to professional audiences without overstating it. The role carries an editorial duty: research published under our name must survive technical review, and a commercial claim must stay inside what an assessment can demonstrate. Writing comes first. Campaign mechanics matter only when they reach the reader who has to make a decision.

What the role covers

These are standing duties rather than a project list. Work is assigned against them, reviewed against them, and used as the basis for the technical conversation at interview.

  1. 01Plan campaigns around a defined audience, the decision that audience must make and a measurement method agreed before launch.
  2. 02Draft articles, briefings and service descriptions with the consultant who delivered the work. Distinguish an observed result, an industry reference and a commercial claim within every sentence.
  3. 03Keep published service descriptions accurate as methods change, and record when each statement was last verified.
  4. 04Coordinate partner relationships, industry events and approved materials, retaining written permission for any third-party content.
  5. 05Evaluate campaign performance without overstating attribution, and state what the available data cannot establish.
  6. 06Obtain written approval before publishing a client identity, assessment finding or confidential detail, and retain the approval record.
  7. 07Read technical material closely enough to ask a useful question of the engineer who wrote it.

What we look for

We look for evidence of each capability rather than a claim to it. Where a requirement cannot be met, say so in the application: a stated gap is easier to assess than an implied one.

  • Business marketing experience in technology or professional services, with evidence of sustained campaigns rather than isolated launches.
  • Professional written English and Traditional Chinese, each composed in its own idiom. Preserving a technical meaning without carrying across a sentence structure is a specific skill.
  • Practical understanding of campaign measurement, including the uncertainty inside an attribution or conversion figure.
  • Interest in cybersecurity sufficient to follow a technical conversation: how organisations commission specialist work, what a control does, why a finding needs evidence.
  • Discipline on sources: quotation, attribution and version dating for any statistic used.
  • A technical writing sample for a business audience, annotated to show which statements would require verification before publication.

How an application is read

Applications are assessed against the duties and capabilities listed for that role, not against a general impression of seniority. Written material decides whether a conversation follows.

  1. 01

    Written material

    A CV with a research sample, portfolio or writing sample, depending on the role. We read for the reasoning behind the result: what was tested, what the method could establish, and what it could not. A claim that cannot be evidenced is read as a claim.

  2. 02

    Technical conversation

    A discussion with the consultant who would supervise the work. Expect to defend a decision you made, to describe a result that did not hold, and to say how you would have found it sooner.

  3. 03

    Decision

    Outcome and reasoning are communicated once the assessment is complete. Where the stated standard is not met, the gap is named rather than left ambiguous.

We value your privacy

We use essential cookies to keep the site working and analytics cookies to understand how you use it, so we can improve. We do not sell your data. See our Privacy Policy for details.