ai ml workflow/launched

YantrikDB

Embedded vector DB with cognitive memory APIs for AI agents.

https://yantrikdb.com

Share

XRedditHNLinkedInFacebook

Instagram has no web-share URL. On mobile use the Share… button — the OS share sheet covers Instagram + WhatsApp + iMessage + others.

verification

L1

Ownership verified · GitHub repo admin: yantrikos/yantrikdb

Ownership verified — not quality endorsement. icantmarket verifies control of listed assets only. last rechecked 9/2/2026.

stars

64

forks

18

created

Feb 2026

pushed

Sep 2026

active ask · launch

Newly published utility patent application covers relevance-conditioned scoring and the cognitive state graph architecture. Looking for one set of eyes — especially anyone who has built vector search at scale or shipped memory for autonomous agents — to read the architecture overview and tell me one thing they would do differently.

Looking for:
Anyone running production AI agents or building vector-search infrastructure. Strong bonus if you have shipped a vector DB or memory system.
Time:
30m
In return:
I will read your memory layer or vector backend with the same lens and send back one concrete suggestion.
Closes:
6/16/2026

Success means: I get one architectural objection that makes me reconsider something I assumed was settled.

Posts · 1

Recent activity · 5

GitHub releases · PyPI downloads · npm versions

  • release

    v0.23.1

    ## v0.23.1 — the embedder you attach is the one that runs Patch release. Schema version unchanged (54): no migration, no heal on upgrade. ### An attached Python embedder now actually runs (#240) - `

    github releases · 9/19/2026, 7:15:00 PM

  • release

    v0.23.0

    ## v0.23.0 — the Memory Atlas is inbuilt, and a terminal explorer Additive release. Schema version unchanged (54): no migration, no heal on upgrade. ### Memory Atlas, packaged (#232) - `yantrikdb/at

    github releases · 9/14/2026, 5:15:00 AM

  • release

    v0.22.0

    ## v0.22.0 — claims earn their authority, and the store refuses a second SQLite library Schema v52 → v54 (`claims.grounding`, `extraction_refusals`). Migration is automatic on open; both heals are id

    github releases · 9/7/2026, 9:15:02 PM

  • release

    v0.21.2

    ## v0.21.2 — a sentence-final name ends at the period; a contraction is never a name Schema unchanged (v52). Two extractor precision fixes (#221), no API change. - **Sentence boundary.** The chunker

    github releases · 9/7/2026, 3:15:02 AM

  • release

    v0.21.1

    ## v0.21.1 — the lexicon writes in one transaction per memory Schema unchanged (v52). One change, no behaviour change (#219). - 0.21.0 folded each memory's case observations into `token_case_stats`

    github releases · 9/6/2026, 9:15:02 PM

Audit log · 5

Public · append-only

  • 9/2/2026, 4:00:02 AMreverifiedtrust=L1
  • 8/3/2026, 4:00:01 AMreverifiedtrust=L1
  • 7/3/2026, 4:00:02 AMreverifiedtrust=L1
  • 6/2/2026, 6:42:20 AMverifiedtrust=L1
  • 6/2/2026, 6:42:20 AMclaimedtrust=L1