Personality Intelligence Platform: MBTI, SaaS Role Fit & Animal Archetype Assessments
The Personality Intelligence Platform is a full-stack application designed for self-discovery and team-fit insights. Users navigate through one of three specialized assessment tracks MBTI-style personality, SaaS role-fit (covering roles like PM, Engineer, Designer), or animal archetype—completing a polished 24-question Likert flow.
The user experience is highly polished, featuring progress tracking, keyboard shortcuts, pause/resume functionality, and a mandatory review step before submission. The backend, built with FastAPI, extracts eight core trait dimensions from the user's answers. It applies a hybrid deterministic scoring engine, combining rule-based trait calculations with optional enhancements from scikit-learn artifacts.
Upon completion, users receive rich, multi-faceted reports detailing summaries, motivations, strengths, growth areas, communication styles, work environment fit, and team role. Each result is assigned a stable `result_code` and a shareable restore link, ensuring report persistence without requiring a heavy database in the MVP.
The Next.js frontend manages the premium landing page, the three distinct test flows, and the styled result pages. It includes an API proxy layer to communicate with the FastAPI backend, and an administrative diagnostics page for monitoring system health, question payloads, and trait mapping.
Features
- Three distinct assessment tracks (MBTI
- SaaS role-fit
- animal archetype)
- Polished
- interactive test flow with pause/resume and review steps
- Rich narrative reports with dimension bars and top-match rankings
- Deterministic result codes and shareable restore links
- Next.js API proxy connecting to FastAPI backend
- Optional ML training pipeline for advanced scoring models
- Admin diagnostics page for system health and validation
Challenges
The primary challenges included the inconsistency of public personality datasets for supervised ML training, the need to balance explainable, stable results with optional ML enhancements, and delivering three distinct, polished assessment experiences within a unified architecture.
Solutions
We implemented a hybrid deterministic scoring engine, using rule-based trait extraction as the core, while treating ML models as optional priors. To manage data complexity, we built an ML pipeline capable of downloading public datasets and generating derived baselines when labels are inconsistent. Architecturally, we cleanly separated the Next.js UI/proxy from the FastAPI prediction/decode API, ensuring a robust and scalable structure.
