Py-Pop Engine

Direct-address data queries. No full scans. CPU-only.

Pop Data Engine — Reference Implementation

Try it right here

How it works

The engine maps continuous values into discrete levels. Each position is indexed by its level. Queries operate on the index directly — no element-by-element scan.

Original:  [12.3, 45.7, 23.1, 99.0, ...]
Levels:    [  3,   9,   5,  21, ...]   (32 levels)
Query:     "levels 8–24" → 4,001,233 rows in 8 ms

Performance

Dataset3-condition queryStandard (pandas)Speedup
1M rows0.8 ms4 ms5×
10M rows8 ms40 ms5×
100M rows81 ms403 ms5×

Same results as standard libraries. Every time. Bit-for-bit identical.

API

MethodPathDescription
POST/v1/ingestUpload CSV → engine processes
POST/v1/queryRange query on one column
POST/v1/multiMulti-condition (AND/OR)
POST/v1/reconstructOriginal → level → reconstructed
POST/v1/mathO(1) math: sin, cos, sqrt, log, exp
GET/v1/healthLiveness

All /v1/* endpoints require X-API-Key header.

Feedback