Zero Gravity Backend

In a Nutshell

A Spring Boot REST API that powers Zero Gravity's emotion logging and AI analysis. The infrastructure and deployment pipeline were built on a free-tier cloud, and the AI features were designed within budget constraints.

From team project to production.

A Spring Boot backend that started as a team project was taken to production. The project originally had only basic CRUD APIs. Authentication, infrastructure, a deployment pipeline, and AI analysis were all added on top.

API Endpoints

Infrastructure

Sending everything would blow the budget in a month.

The goal was to build a feature that analyzes emotion records via the Gemini API. Sending a full year of records meant ~55K input tokens per request. That was not sustainable on the project budget.

Reducing data would hurt analysis quality, but dropping the feature was not an option since it was core functionality.

One bucket, one representative record.

If sending everything was off the table, the solution was to pick the most representative record from each time period. The period was split into equal time units (buckets), and one representative record was taken per bucket. A Year analysis means monthly units, so twelve of them. Picking from the whole set at once would cluster them in the months with the most records, losing the shape of the year.

That left the question of how to define "representative." The criteria were already inside the service: the average emotion level and reason statistics the chart aggregated per period.

The reason statistics were re-aggregated per bucket, and the aggregation queries built for the emotion statistics were reused as they were.

Scored at 60% level, 40% reason.

Each bucket's representative was selected by a weighted score of emotion level 60% + reason match 40%.

private double calculateMatchScore(
    EmotionRecord record, Double targetLevel, String topReason
) {
    // Levels run 0-6, so the maximum distance is 6
    double levelScore = 1.0 - (Math.abs(record.getEmotionId() - targetLevel) / 6.0);
 
    double reasonScore = record.getEmotionReasons().contains(topReason) ? 1.0 : 0.0;
 
    return (levelScore * 0.6) + (reasonScore * 0.4);
}

Ties were broken by longest diary first, most reasons, then most recent.

Example: January bucket in a Year analysis (avg emotion level: 4.5, top reason: "Work")

97% payload reduction, $0.002 per request

Note: AI analysis results were cached for 24 hours, and the cache was invalidated for the relevant period whenever emotion records changed.