← Andrew Crenshaw Strata
Autonomous Job-Search OS

Strata

Treat job search as a strategic campaign, not a checkbox exercise.

What it is

Strata is an autonomous AI job-search operating system. Instead of scrolling job boards, you define your profile and target companies, then Strata's agents handle the rest: scanning career pages daily, detecting new listings, scoring them against your profile, and surfacing the matches worth your attention, with explanations of why each role matches and where the gaps are.

Every application requires your approval. Strata optimizes for signal quality, not spray-and-pray volume.

Design principles

PrincipleWhat it means
Precision over volumeEvery application requires human approval. The system surfaces fewer, better opportunities, not more noise.
Agent autonomy within governanceEach agent has access to only what it needs. The Scraper Agent can't modify your profile. The Profile Agent can't submit applications. Enforced at the database level, not by convention.
Cost-conscious by defaultA three-layer deduplication funnel eliminates 95%+ of duplicate listings before any expensive AI scoring happens.
Observable alwaysEvery agent decision is logged. Ask "why did this job score 82?" and get the full reasoning chain.

The eight agents

AgentWhat it doesYour involvement
ProfileConverts your resume into a rich Candidate Intelligence Profile (CIP), with full version historyYou review and approve all identity data
TargetManages your target company list with priority tiers (Tier 1 to Tier 3)You define your strategy and company set
ScraperDaily career-page sweeps across all your target companiesRuns automatically; auto-promotes validated results
MatchTwo-stage AI scoring: vector similarity first, then deep LLM analysis for borderline casesRuns automatically; surfaces results with confidence tiers
ApplyGenerates tailored resumes and cover letters for high-match rolesYou always review before anything is sent
SkillsGap analysis against your highest-scoring matches: what skills would unlock more opportunitiesRuns automatically from active match data
NetworkContact intelligence and outreach-timing recommendationsYou verify imported contacts
BrandLinkedIn content and portfolio positioning aligned to your target rolesYou approve all published content

How it works

Your profile + target companies ↓ Daily career-page scrapes (50+ ATS platforms supported) ↓ Change detection: new listings, re-posts, de-escalations ↓ 3-layer deduplication (URL → content hash → vector similarity) ↓ AI matching with full reasoning traces ↓ Your dashboard: scored matches, explanations, action queue ↓ Application materials (resume + cover letter): you click send

What makes it different

You control every decision

Strata never auto-submits applications, never modifies your profile without approval, and never sends anything on your behalf. Governance is enforced at the infrastructure level.

The system explains itself

Every match score includes a reasoning trace: which requirements aligned, which gaps exist, and what's missing from your profile. No black boxes.

Signal over noise

Most job-search tools optimize for volume. Strata optimizes for relevance. The dedup pipeline means you see each opportunity once; scoring tiers separate strong matches from weak ones before they reach your queue.

Efficient by design

The three-layer deduplication funnel filters the vast majority of listings before any AI call. You can also run it fully local via Ollama.

Status

Alpha. The core discovery and matching pipeline is functional and in active use. The Next.js dashboard handles profile management, job review, target tracking, and application management. Notification channels and public release are in progress.

Open-source libraries

Two components of Strata are available as standalone Python packages, stateless and MIT-licensed. They return structured data; Strata decides what to persist and how to govern it.

PackagePurposeInstall
strata-harvestCareer-page scraping across 50+ ATS providerspip install strata-harvest
strata-matchTwo-stage vector + LLM job-to-profile matchingpip install strata-match

The Strata core application is proprietary. The strata-harvest and strata-match libraries are open-source and available on PyPI.