Blog
Technical deep dives on health insurance data, API design, and building reliable data infrastructure.
How We Paginate 22,000 Health Plans: Cursor, Offset, and When to Use Each
A technical deep-dive into Opelyx's dual-mode pagination — cursor-based keyset pagination for data pipelines and offset pagination for browsing — with real D1/SQLite numbers, compound cursor encoding, and the honest tradeoffs of each approach.
Health Insurance Data Quality: What We Learned Processing 22,000 Plans
Real data quality failures from processing CMS Public Use Files and state exchange data — swapped URL columns, junk brochure links, HIOS ID mismatches, and why we store everything in R2 before touching it.
SBC Data Explained: Making Out-of-Pocket Costs Transparent
Summary of Benefits and Coverage documents contain standardized cost scenarios that reveal what a health plan actually costs beyond the premium. Here's how we extract and use this data.
The State-Based Marketplace Data Challenge
21 states run their own health insurance exchanges and each publishes plan data differently. Here's what we learned building a unified parser that handles all of them.
Building a Health Plan Comparison API: Architecture Decisions
How we built the Opelyx Health Plans API on Cloudflare Workers, D1, and Hono — and why each of those choices was the right one for sub-50ms global response times.
How CMS Public Use Files Power Health Plan Data
A technical deep dive into the CMS Public Use Files that underpin ACA marketplace plan data — what they contain, when they're released, and the engineering challenges of processing 1.4 million rate records.
Understanding ACA Rating Areas: A Developer's Guide
Rating areas govern how ACA health plan premiums are set by geography. If you're building with health plan data, getting rating area lookup wrong will silently return incorrect prices.
Why We Built Opelyx: Structured Data for Health Insurance
Health insurance data is a mess — fragmented CMS files, inconsistent formats, and API vendors who won't talk to you unless you have a $10K budget. Here's why we built a developer-first alternative.