iBola is a production agentic RAG chatbot that answers questions about Bolaji’s experience, skills, projects, and availability. It is the interactive way to explore this site: ask it anything a recruiter or client would ask.

Architecture: a LangGraph workflow (guardrail, retrieve, grade, generate, verify) with Pydantic structured outputs, hybrid retrieval (BM25 + vector + reciprocal rank fusion), and claim-level grounding so every factual answer carries evidence from the knowledge base.
Knowledge base: ingested from this site’s canonical machine-readable files, so the chatbot inherits every site update automatically.
Quality: evaluated on a 200-question bilingual golden set (EN/FR, adversarial and multi-turn included) with an LLM-as-judge harness on Langfuse; canonical-fact accuracy is gated at 100% in CI before any deploy.
Operations: Google Cloud Run with canary deploys, per-node tracing, user-feedback scores, and a monthly assessment pipeline.

Ask iBola →

iBola.