JobClock — Full-Stack Job Search SaaS

Engineering highlights

1. I

   Built a full-stack AI-powered job search platform that aggregates 7 job sources through a unified search API, integrating Adzuna, Reed, CareerJet, Greenhouse, Lever, Ashby, and Workday.

2. II

   Designed a fault-tolerant aggregation pipeline using Promise.allSettled, ensuring upstream API failures degrade individual results rather than the entire search experience.

3. III

   Implemented source-specific query optimisation and cross-source deduplication by normalising job titles and prioritising canonical ATS listings over syndicated duplicates.

4. IV

   Developed a 4-stage AI CV tailoring pipeline (job analysis, CV matching, tailoring strategy, CV generation) using OpenAI and Anthropic with Zod schema validation and automatic response repair for reliable structured outputs.

5. V

   Built a Chrome Manifest V3 extension that extracts job descriptions from major job boards using DOM-aware parsing with structured data fallbacks (Open Graph and JSON-LD) for high extraction accuracy.

6. VI

   Implemented secure extension authentication using rotating single-use API tokens stored as SHA-256 hashes with revocation support.

7. VII

   Architected asynchronous AI processing with Upstash QStash, supporting idempotent background jobs, job deduplication, per-user rate limiting, and local development fallbacks.

8. VIII

   Secured user data with PostgreSQL Row-Level Security, AES-256-GCM encryption for API keys, and multi-layer sliding-window rate limiting across application endpoints.