Hi HN — I built SEE, a schema-aware searchable compression format for JSON/NDJSON. The goal isn’t just smaller bytes; it’s reducing the hidden “data tax” by enabling exists/pos-style lookups while compressed (page-level random access + high skip rates).

KPI snapshot (reproducible):

Combined size ratio: 7.7% (SEE/RAW) vs 13.7% (Zstd/RAW)

Lookup latency (cold, ms): present p50 0.083, absent p50 0.084

Skip: present 0.994, absent 0.997

Proof-first evaluation (no meetings required): offline 10-min demo, DD evidence pack, and a BYOD kit to test on your own NDJSON locally.

Flippa listing (acquisition / exclusive license): https://flippa.com/12431275

Evaluation hub: https://gitlab.com/kodomonocch1/see_proto

I’d love feedback on where you’d integrate this (object storage archives, logs/observability, data lake scan workloads, etc.).