Selected work.

Problem

Job search is noisy, manual, and impossible to measure

Candidates juggle dozens of job boards with inconsistent data, low-trust companies, stale postings, and no feedback loop between applications and outcomes. Recruiting teams face the mirror image: high-volume pipelines with weak signal and no shared system of record.

Constraints

Recruiter-safe by design

The platform had to operate inside real recruiting norms — no shortcuts that would compromise safety, trust, or data quality.

• No CAPTCHA, MFA, login-wall, or bot-detection bypass
• Manual review fallback on every automated decision
• Salary, freshness, and source-confidence are first-class signals
• ATS differences across Greenhouse, Lever, Ashby, and Breezy
• Daily caps and emergency pause built into the workflow

Solution

An AI-powered career operations platform

CommandFlowOS combines verified job intelligence, resume optimization, recruiter CRM workflows, outcome learning, and safety-aware automation into a single product — with a public sandbox for storytelling and a private workspace for real operations.

• ATS ingestion across Greenhouse, Lever, Ashby, Breezy
• 11-factor company trust score and scam-risk detection
• Resume Variant Optimizer with A/B testing across 5 canonical variants
• Recruiter CRM pipeline from Imported → Offer
• Outcome Learning Engine that recalibrates scoring from real recruiter signals
• Executive Command Center for recruiter-safe KPIs

Product Decisions

Why the system is built this way

Each layer exists for a specific reason — to keep the platform trustworthy, measurable, and useful to both candidates and recruiters.

• Separate public sandbox from private workspace so storytelling never leaks real data
• Score trust before matching so weak companies never reach the pipeline
• Treat outcomes as training data, not vanity metrics
• Make every automation reversible with a manual fallback
• Use a recruiter-style CRM model so the workflow feels familiar to hiring teams

Impact

Measurable outcomes, not buzzwords

• 1,441+ raw jobs tested across multiple ATS sources
• 460+ jobs filtered and imported into the recruiter pipeline
• 11-factor company trust scoring with a 5.4% Auto Review pass rate
• 5 canonical resume variants with outcome-driven selection
• 14-day freshness filter and 30-day duplicate-company guard
• Executive Command Center surfaces KPIs that previously lived in spreadsheets
• Manual review effort focused on the top of the funnel, not every job

Roadmap

What improves next

The next phase deepens the learning loop and makes the platform easier for recruiters to adopt as a system of record.

• Recruiter-side intake — invite hiring teams into the CRM with scoped roles
• Calibrated interview-probability model from real outcomes
• Industry-specific trust models (startup vs. enterprise vs. agency)
• Higher-resolution outcome attribution per resume variant and ATS source
• Public benchmark reports drawn from sandbox data only