All work

QuestBoard

Job discovery, resume preparation, application actions, contact research, outreach drafts, replies, and statuses can become disconnected, creating duplicate work and unsafe automation.

Role
Product Engineer
Type and status
Private tool · Active development
Period
2026

See the system.

Actual interface evidence is embedded here. External links are secondary.

QuestBoard desktop job detail using its fictional Test Example.
Synthetic demo data

Materials, application state, contacts, unsent drafts, replies, notes, and history stay in one record.

QuestBoard Test Example capture, August 24, 2026
QuestBoardJob operating record
QuestBoard desktop job detail using its fictional Test Example.

Materials, application state, contacts, unsent drafts, replies, notes, and history stay in one record.

The operational problem

Job discovery, resume preparation, application actions, contact research, outreach drafts, replies, and statuses can become disconnected, creating duplicate work and unsafe automation.

A private job-operations CRM that keeps opportunity state, materials, actions, contact evidence, replies, and human review around one candidate workflow.

Why this system exists

One candidate-scoped system can preserve evidence and lifecycle state while external workers remain bounded and human decisions stay explicit.

External workers may discover, score, research, or prepare. QuestBoard stores durable state, enforces safety boundaries, and presents the review workflow.

Who uses itA separate candidate named Ali. Ayub built the software; he is not presented as the candidate using this instance.

How the work moves

  1. Import job
  2. Review fit and duplicate risk
  3. Prepare job-bound materials
  4. Review application action
  5. Record confirmed submission
  6. Unlock contact review
  7. Track replies and outcomes

What shaped the solution

The current browser serves a separate candidate named Ali.

Application submission and outreach remain policy-gated and human-reviewed.

What I owned

Designed

Candidate-scoped state, lifecycle derivation, safety gates, and review surfaces

Developed

Home, jobs, detail, actions, stats, settings, documents, notes, and timeline

Integrated

Turso, signed documents, read-only inbox sync, and external worker contracts

Tested

Migrations, idempotency, hostile markup, policy gates, uncertain submission, caching, and responsive UI

Context matters. The work happened with teams and stakeholders. These statements describe my contribution without turning shared delivery into a solo claim.

Where AI fits

Model or agent role

External workers can discover jobs, analyze fit, tailor resumes, research public hiring contacts, and draft review material.

Structured contract

Workers write versioned evidence through authenticated, idempotent contracts.

Deterministic safeguards

Candidate policy, exact job binding, duplicate checks, approved answer keys, provider fingerprints, signed documents, and confirmation evidence prevent unsupported actions.

Human decision

The current candidate policy is incomplete, so submission and outreach remain manual or stop at review-ready state.

Decisions that carried the work

Make the CRM the safety gate

External workers can prepare evidence, but durable status changes require authenticated, idempotent records and confirmation.

Verified outcome

Candidate-scoped job, document, application, contact, message, inbox, and timeline state.

Deprecated sending and discovery routes fail closed.

Responsive checks cover mobile through wide desktop.

What I will not overclaim

The portfolio uses only the explicit fictional Test Example. It does not present Ali's job search as Ayub's own or claim autonomous submission and outreach.

Verified from the current QuestBoard repository, Ayub-authored history, README, PRODUCT, local demo runtime, and Test Example screenshots.

Tools used where they fit

Node.jsJavaScriptSQLiteTursoHTMLCSSPlaywright

Continue through the work

Que Media WebsiteEmail Ayub