Keyrxng

Semantic Task Matchmaking: Legacy Vector Gravity After 6+ Months Away

8/13/2025 · 7 min

So what?
Shows enduring cross-domain impact while proposing concrete scoring upgrades to reduce historical bias.

See the related case study: Command-Ask: Context-aware GitHub AI Assistant

Series linkage: This field note is the catalyst that led to Part I (diagnosis) and Part II (playbook) of the “Ghost in the Machine” series. It captures the raw observation—persistence without presence—and frames the narrative for stakeholders. See Glossary for terminology.

1. Context: the triggering observation

During my tenure the company (Ubiquity) introduced an AI + RAG semantic task matchmaking layer: vector embeddings of past work, completed issues and tasks, driving “suggested / tagged” developer assignments atop Ubiquity OS. I have been absent >6 months. Yet in the latest metrics export (self-generated 2025-08-19) I remain the second most semantically tagged contributor.

Source table: contributor_table.csv (timestamp: 2025-08-19T22:21:54.761Z). Columns: Merged PRs, Semantic Tags, Dormant windows, etc.

2. Snapshot: tag distribution (evidence)

RankContributorSemantic TagsMerged PRsTags / Merged PR
1gentlementlegen1977830.25
2Keyrxng741360.54
3whilefoo471190.39
40x4007382830.13
5shiv81035550.64
6ishowvel20141.43*

*Small denominator; high ratio but low absolute surface.

Condensed

Drivers, remediation formulas, and metrics now live in Ghost series Part I & II; this field note retains only the trigger and tag table.


Legacy vector gravity reveals temporal resistance—this note records the spark.

See also