← All work
SST Oracle
AI candidate search and talent intelligence for African hiring teams
SST Oracle is the product side of Silicon Savannah Talent: a semantic candidate search platform that lets recruiters and hiring managers find African talent that keyword search misses.
I am co-founder and lead technology architect, not the sole owner.
- 840
- automated tests
- 4
- role-based apps
- Escrow
- milestone releases with splits
Why
African talent is often invisible to global applicant-tracking systems. Most platforms index keywords, not skills, and miss candidates whose CVs do not match a Silicon Valley vocabulary. Recruiters end up trusting their networks because the tools do not earn it.
What it does
- PostgreSQL with pgvector, embeddings stored beside structured CV data.
- Gemini embeddings for queries and candidate profiles, re-embedded as the model improves.
- AI CV parsing and job scoring; recruiter-grade ranking that blends semantic similarity, structured filters and recency.
- Milestone escrow with per-member payout splits, so teams get paid when the work lands.
- OTP-only authentication, four role-based apps (candidate, client, admin, API), n8n for ingestion and re-scoring jobs.
What I led
- Overall technology architecture and stack selection.
- Embedding strategy and the retrieval pipeline.
- A live database audit that found real schema and pipeline failures a green test suite had missed.
- Engineering hiring and delivery cadence with my co-founders.