Talero AI
Co-founder of a small startup that turns career assessment into a story-driven game, with dashboards, a university programme and an AI hiring platform around it.
Talero AI turns career assessment into a story-driven game that helps people find their strengths and where they fit, and builds products around it for individuals, universities and employers.
Talero has two founders. I am responsible for the technology across every product; my co-founder leads marketing and sales.
1,000+ people have played the game, and the first paying university partner in Germany has 200+ students on it.
Visit taleroai.com
The project
A small startup needs one technical owner who can keep a game, several dashboards, an AI hiring platform and the backend they share working as one system. That is my role at Talero.
What I built
- Talero Wise: a story-based assessment game (Unity/WebGL) with Viking, Samurai and Roman worlds. I own its technical decisions and the backend it runs on, and built features such as authentication.
- Talero Talents: a dashboard that turns game results into personal career reports, with a paid plan and a portal for career coaches.
- A university programme: its own game build, a student dashboard and an advisor dashboard that advisors use to prepare one-to-one study-orientation conversations.
- Neuro-Blitz: a mobile orientation game for the same university partner.
- Talero ATS: the AI hiring platform I built.
- One shared Node.js and TypeScript backend with a single sign-in across every game and dashboard, running on Google Cloud.
- The public website at taleroai.com, designed and built in React and TypeScript: GTmetrix grade A with 99% performance, and readable by AI agents as well as people.
A pixel-art website that loads fast
I designed and built taleroai.com, where players, universities and employers first meet Talero. It carries the game's pixel-art worlds and still loads fast: in GTmetrix's October 2026 test it scored grade A, with 99% performance, 96% structure, a largest contentful paint of 632 ms, 27 ms of blocking time and no layout shift. An agent-readiness scan rated it 80, Level 5 "Agent-Native", with full marks for discoverability, content and bot access control. API, sign-in and MCP discovery for agents is the part still to finish, at 5 of 8 checks.
A result screen that always said the same thing
Neuro-Blitz ends by suggesting three study programmes. An audit of its result engine showed that, whatever players chose, the ending kept offering the same three. The scoring itself worked; the step that turned scores into programmes did not. The matching logic was rebuilt so the suggestions follow what each player actually did.
One identity across every product
Players sign in once with Google, and the same account carries their results from the game to their dashboard. The regular game and the university build share one authentication project and one backend, so a new game build or dashboard plugs into the same API instead of growing its own copy.