Engineering
9 min readDeterministic extraction from financial PDFs without the hallucinations
How we replaced a chain of LLM calls with a small, typed extraction pipeline — and shipped 99.4% field accuracy on 10-K filings.
DamanFounding Engineer
Blog
Engineering deep-dives, research, and product changelogs from the HyperAPI team.
Research
7 min readSchema is the API contract — and the prompt
Why we treat customer-supplied JSON schemas as the source of truth for parsing, classification, and validation in a single round-trip.
Daman
Engineering
11 min readClassifying the long tail of bank statement formats
A field guide to handling 1,400+ statement templates without a per-bank model. Spoiler: the trick is in the layout features, not the text.
Daman
Product
4 min readIntroducing the Classify API
One endpoint. Any document. Returns a typed label plus a calibrated confidence — generally available today.
Daman
Tutorial
12 min readBuilding a loan-underwriting pipeline with the Extract API
A walkthrough: from a stack of borrower PDFs to a structured underwriting dataset in under 60 lines of TypeScript.
Daman
Changelog
3 min readChangelog · April 2026
Streaming Extract responses, Python SDK v2, sub-second cold starts on Parse, and a quieter dashboard.
Daman