Mentor IA

Founder & Lead Engineer

An AI mentoring system focused on dependable system design, not one-off model demos.

The project is structured as a production-oriented case study: clear API boundaries, retrieval-backed responses, and operational feedback loops for continuous improvement.

Instead of a feature-first narrative, the implementation prioritizes architecture choices that support reliability, maintainability, and future scale.

FastAPI MongoDB RAG TypeScript
Mentor IA project overview

Problem

Students often receive inconsistent guidance across fragmented channels, which makes it hard to maintain momentum and confidence during demanding coursework.

Solution

Mentor IA delivers contextual guidance through a backend that combines retrieval, orchestration, and response policies. The core design separates interaction logic from model logic so each layer can evolve independently.

The result is a system-focused implementation where user state, knowledge retrieval, and response generation are coordinated through explicit service boundaries and versioned prompts.

System Architecture

Mentor IA system architecture diagram

Data flow: client request → API orchestration layer → retrieval/context assembly → LLM response generation → persistence and telemetry logging.

Engineering Decisions

Validation & Iteration Strategy

Validation currently relies on structured qualitative review and controlled test scenarios rather than production KPI reporting. Each iteration focuses on response usefulness, retrieval relevance, and failure-mode handling.

As the system matures, evaluation will track response quality trends, retrieval precision, user progression signals, and operational stability indicators through a repeatable review cadence.

My Role

View Code Back to Projects GitHub Demo