← Back to Blog
·6 min read

Every Student Project Deserves a Decision Log

StudentsPortfolioLearning

You built something genuinely impressive over the weekend. A full-stack task management app with real-time updates, OAuth login, and a clean UI. Claude Code helped you scaffold the whole thing. You pushed it to GitHub, added a README with screenshots, and pinned it to your profile. Portfolio piece: done.

Then the interview happens.

"Nice project. Why did you choose PostgreSQL over MongoDB?"

You pause. The honest answer is: Claude suggested it and it worked. But you can't say that. So you mumble something about relational data and ACID compliance — words you half-remember from a database course — and hope the interviewer moves on.

"What made you pick Next.js instead of a Vite SPA?"

Another pause. You genuinely don't remember. It was two weeks ago and you were moving fast. The decision happened in a conversation that no longer exists.

The interviewer isn't trying to catch you. They want to know if you can think about software, not just produce it. And right now, your project shows code but hides thinking.

The Portfolio Gap

Every CS student and bootcamp grad in 2026 has GitHub repos. Many of them were built with AI assistance. Hiring managers know this. The repos themselves are no longer differentiators — what separates candidates is evidence of engineering judgment.

A GitHub repo shows what you built. It shows your code, your file structure, your commit history. What it doesn't show is why. Why this database? Why this framework? Why this architecture pattern? Why did you refactor the auth module halfway through? What alternatives did you consider and reject?

This is the portfolio gap. The most valuable signal a junior developer can send isn't "I can write code" — every new grad can write code, especially with AI. The signal is: "I can make deliberate technical decisions and explain my reasoning."

A decision log closes that gap. It turns your project from a static artifact into a narrative of engineering thinking.

What a Decision Log Looks Like

A decision log isn't an essay. It's a structured record of choices. Each entry has three parts: context (what situation prompted the decision), decision (what you chose), and consequences (what this means going forward).

Here's a real example:

Decision: Use Fastify over Express for the API server

Context: Need a Node.js HTTP framework. Team is two people, both comfortable with TypeScript. Performance matters because we're handling real-time WebSocket connections alongside REST.

Alternatives considered: Express (larger ecosystem but slower, weaker TS support), Hono (very fast but young ecosystem, fewer middleware options for our use case).

Consequences: TypeScript-first with built-in schema validation. Smaller middleware ecosystem than Express, but sufficient. Plugin system makes code organization cleaner.

Read that entry and you immediately understand the reasoning. An interviewer reading this sees a developer who evaluates tradeoffs, considers alternatives, and thinks about consequences. That's exactly the signal they're looking for.

Even simple decisions matter. "Chose Tailwind CSS over plain CSS because rapid prototyping speed matters more than bundle size optimization for this project." One sentence. Shows prioritization and awareness of tradeoffs.

Capture While You Code

The reason most developers don't keep decision logs is friction. You'd have to stop coding, open a separate document, write up the decision, and go back to coding. Nobody does that during a flow state. By the time you remember to write it down, you've forgotten the reasoning.

Sprintra solves this by embedding decision capture directly in your AI coding session. When you're working with Claude Code and you make a technical choice, you just say:

"Record a decision: we're using PostgreSQL instead of MongoDB because we need relational joins for the reporting feature and JSONB gives us document flexibility where we need it. Considered MongoDB but the aggregation pipeline is harder to maintain."

That's it. The AI creates a structured Architecture Decision Record with context, decision, alternatives, and consequences — all extracted from the conversation you're already having. No context switching. No separate tool. No friction.

Over the course of a weekend project, you'll naturally accumulate 5-15 decisions. Database choice, framework selection, authentication strategy, deployment platform, state management approach, API design patterns. Each one captured in real time with the full reasoning intact.

From Weekend Project to Interview Story

A project with a decision log tells a story. It transforms a GitHub repo from "here's some code I wrote" into "here's how I approached a problem."

But decisions are just one piece. A fully structured project also includes:

  • Features with acceptance criteria — shows you can define scope and requirements
  • Stories with status tracking — shows you can break work into manageable pieces
  • Sprint planning — shows you can prioritize and set realistic goals
  • Session notes — shows your thought process as it evolved

Imagine walking into an interview and saying: "Here's my project. I planned five features, broke them into twenty-two stories, made twelve architecture decisions, and shipped it in a weekend. Here's the decision log. Here's why I chose every major technology. Here's what I'd change if I built it again."

That's not a junior developer talking. That's an engineer. And the difference is a decision log that took zero extra effort to maintain because it was captured during the conversations you were already having.

There's a bonus benefit: you actually learn more. When you record why you chose PostgreSQL, you internalize the reasoning. Next time someone asks — in an interview, in a code review, in a team discussion — you don't have to reconstruct the logic. It's already part of how you think about the problem.

Free to Start

Sprintra is free for individual developers. Two ways to get started:

Self-hosted (completely free): Run npx create-sprintra and you're set up in under two minutes. Local SQLite database. Works with Claude Code, Cursor, and Claude Desktop via MCP. Your data stays on your machine.

Cloud (free tier): Sign up at app.sprintra.io. Same features, accessible from anywhere. Great if you work across multiple machines.

Your next weekend project is going to produce great code either way — AI makes sure of that. The question is whether it also produces evidence that you're a thoughtful engineer. A decision log is the difference between "I built this" and "I built this, and here's why."

Build a portfolio that shows your thinking

Capture decisions, track features, and turn weekend projects into interview stories. Free for students and solo developers.

Get started →