Loading project case study
All work

Personal Projects / 2026

Competency Manager

Connected expectations, assessment, growth, and hiring in one competency system so designers receive clearer feedback and leaders can invest against visible team needs.

Visit live app Read the case study
Competency Manager

Value and impact

One source for expectations
Spreadsheets → Unified System
Evidence-based development
AI-Generated from Real Data
Hiring reflects team needs
Team Gap-Informed

01

Overview

Competency Manager is a full-stack web application I designed and built to bring structure and consistency to how I evaluate, develop, and hire Product Designers.

It replaces scattered spreadsheets and ad-hoc processes with a single system where one competency framework powers everything: team assessments, trend tracking, hiring pipelines, skill mapping, and AI-generated development plans. The primary user is me, a design manager running a team of product designers across multiple levels, from Associate to Principal. Team members have viewer access to their own assessments and progression plans.

  • Product Design
  • AI/ML
  • Full-Stack
  • Management Tools

Challenge

Managing a design team well requires making a lot of judgment calls, and most of the tools available to support those decisions are disconnected from each other.

Evaluation is subjective and inconsistent without a shared rubric. Assessment and hiring are separate worlds with no way to connect what the team needs with what a candidate brings. Growth plans are disconnected from actual assessment data. Team skill gaps are invisible until you're already struggling in a project. Existing tools, spreadsheets, Notion databases, HR tools, each solved a piece of the problem but none connected assessment data to hiring decisions to development plans in a coherent way.

Approach

Started with a core design principle: one competency framework should be the backbone of everything.

If the same criteria define what 'good' looks like at every level, then assessments, hiring, skill mapping, and growth plans can all draw from the same source of truth. Integrated Claude to generate promotion plans, assessment summaries, interview questions, and hiring recommendations, always grounded in actual assessment data. Used Convex's reactive data model so every rating is instantly persisted and reflected with no save buttons or stale state. Designed for progressive complexity, simple surface, deeper insights as you go, with a tool-like, dark-mode aesthetic reflecting its nature as a management utility.

04

Solution

Built four interconnected modules on a single competency framework.

The Competency Framework defines sub-competencies with level-specific criteria across IC and management tracks, with drag-and-drop reordering and Markdown/JSON export. Team Assessments use a wizard-based flow with 5-point scale ratings, auto-calculated scores, trend charts across cycles, and AI-generated summaries and discussion prompts. Team Skill Mapping aggregates assessments into radar charts showing collective capability and surfaces gaps with hiring recommendations. The Hiring Pipeline evaluates candidates through configurable multi-stage pipelines scored against the same framework, with AI-generated interview questions, signal guidance, and team-fit assessments.

05

Impact

Designers can see what is expected at their level and receive feedback tied to specific criteria rather than general impressions.

Managers can connect assessment trends to actionable development plans, while hiring teams can evaluate candidates against both role expectations and visible capability gaps. For the design organization, one framework replaces disconnected spreadsheets and makes growth, staffing, and hiring decisions more consistent and explainable.

06

Key learnings

Designing for yourself is a double-edged sword, deep domain knowledge lets you skip user research and move fast, but scope creep is constant.

AI works best when constrained by real data, structured assessment data fed into tightly scoped prompts produces far better output than open-ended prompts. Real-time data changes the UX, Convex's reactive model made assessment flows feel like continuous interactions rather than form submissions. The framework is the product, making the competency framework the single data model that everything composes on simplified the architecture and made every new feature automatically coherent.

PreviousLiftroom View all projects