KODE WITH KLOSSY × CODEX

A home for Klossy alumni

Role

Product Manager & Engineer

Timeline

June 2026 - July 2026

Team

3 People

Tech Stack

React, TypeScript, Tailwind
Next.js
Drizzle ORM, PostgreSQL

As a fellow in Kode With Klossy's AI BuildLab, sponsored by Codex, I served as product manager and engineer on a team of three, ideating and shipping a product in six weeks. I ran user research and roadmapping, then directed AI coding agents like a technical lead, coding and debugging across the full stack. The result, Kircuit, is a real home base for Klossy alumni across all of their programs.

Kode with Klossy alumni lose touch with each other and the organization after their summer ends.

There's no shared space to keep collaborating, and organizers have no single tool to run community programming beyond emails. As program alumni ourselves, we experienced this fragmentation firsthand.

The Kircuit: A community platform centralizing events, networking, mentorship, and communication for the KWK community

Community Hub landing page showing interest-based communities like AI for Artists, Indie Game Developers, and Local Climate Action

Build community around shared interests

Resources panel with a searchable list of community-shared links

Share resources and opportunities

Events panel showing upcoming Kode with Klossy events near you

Find events near you

Alumni network directory for connecting with other Kode with Klossy alumni

Connect with other Kode with Klossy Alumni

I interviewed users across the Kode with Klossy community. Through 20–30 minute conversations, I uncovered their needs, pain points, and goals. Scroll to see what I learned!

I used these findings to develop two user stories.

User Story #1: Maya

Maya is a 20-year-old college student who attended KWK camp three years ago. She still thinks fondly of the friends she made in that cohort, but outside of a few scattered group chats, she's lost touch with most of them and has no idea what the newer cohorts are working on. When she logs into the Hub, she finds a community for alumni in her city, posts a question about a career move, and gets three replies within a day.

User Story #2: Priya

Priya is a KWK program organizer running community engagement for 200+ alumni across five cities. Before the Hub, she managed events and resource-sharing through a patchwork of spreadsheets, email blasts, and Instagram stories that not all alumni saw. Now she creates an event directly in the Hub, and it's instantly visible to every member of the relevant community—no more chasing RSVPs by hand. She spends less time on logistics and more time actually building the connections and support alumni need.

I turned my interview insights into product requirements that prioritized connection and discoverability.

Product Need Key Requirements
Build Communities Create public interest-based communities, discover communities from the homepage, and join groups.
Enable Discussion Create posts, ask questions, reply within communities, and participate in discussions.
Share Resources Upload articles, opportunities, and tutorials, and search resources within communities.
Support Events View upcoming events with dates, times, locations, and descriptions, and allow organizers to manage community events.
Member Directory Browse member profiles, search members by name or interests, and connect with alumni, peers, and mentors.
Hand-drawn journal sketch wireframing The Kircuit's home, community, and member directory pages

Prototyping in my journal

Annotated Figma wireframe for The Kircuit homepage, including navigation, communities, events, people, resources, and footer sections

Wireframing in Figma

Let's develop this thing!

FRONTEND: React, TypeScript, Tailwind BACKEND: Next.js DATABASE: Drizzle ORM, PostgreSQL. TESTING: Jest

Through BuildLab, I developed a strong foundation in AI-first development, learning best practices for prompting, testing, and tracing feature workflows. In this example, when an alumna shares a resource with her community, the request flows through the full stack:

GIF demoing a user adding a resource in Kircuit
  1. Open form

    The community page renders NewResourceForm, a client-side React component.

  2. Submit

    On submission, the form converts its data into JSON and sends a POST request to the /api/resources route handler.

  3. Validate

    Try-catch blocks validate required fields, ensure the URL is valid, and query the database to verify the specified community exists.

  4. Save

    Drizzle ORM inserts the resource into the PostgreSQL "resources" table.

  5. Refresh

    router.refresh() reloads the server data to display the newly added resource.

To deliver a usable MVP within our timeline, I focused on solving the community's most critical needs before expanding into engagement features.

Depth > Breadth

Tradeoff

I prioritized publishing posts, sharing resources, and organizing events over features like notifications, private messaging, and recommendation algorithms.

Impact

This kept the MVP focused and let us validate the community's core value proposition before investing in additional social features as the user base grows.

Here's what I'm taking away.

Constraints make a product better, not worse.

Every feature I cut didn't have a clear enough scope to survive my initial vision for the product. Learning to say no early forced me to interrogate what our users actually needed and identify what features would really add value.

Product and engineering are not separate disciplines.

Writing the PRD and prompting the AI coding agent were part of the same job, not separate ones. Knowing what to build and how to scope it well mattered just as much as knowing how to code it.