matchmaking-v2

growqr-app/matchmaking-v2

Fork 0

Commit Graph

Author	SHA1	Message	Date
raulgupta	be921559e7	Real Scout dashboard backend + per-board cursor rework - get_scout_stats aggregator (app/engine/stats): assembles the REAL Summary metrics — funnel + cohort engagement rank + active-window (activity timestamps by hour) from our DB, accumulating salary band (avg of each deck's peak, ₹L), match/competition stats from the feed, Momentum/QX + Q-Score trend (qscore-service), day streak (user-service). Honest: unsourced cards return None so the UI omits/locks them, never faked. posted_date extractor. - Activity tracking: viewed/saved flags + search_count → funnel (Matches→Viewed→Shortlisted→ Applied) + engagement percentile. matchesFound = all-time count. - Per-board search cursors {board: page} (replaces the single cursor): only boards that truly paginate (LinkedIn) get one; cursor = LAST page fetched (1st search of a new query → 1). Resets on query change OR >24h (boards refresh ~daily). Dropped Naukri incremental/stateKey (opaque, exhausting, cross-account dedup state) — dedup is the PER-USER seen-net only. - tests: stats helpers (posting-age, histogram, active-window, engagement score).	2026-06-21 12:12:28 +05:30
raulgupta	b1e9cdd182	Engine v2: embeddings sift + Opus curator + Postgres feed persistence - A1: fire the hard filter (Apify is precise); embedding vibe-ranker (embed.py) blends with the floor-free white-box (rank.py) to sift ~90 -> top 18 - A2: Opus curator (curate.py) reads the 18 -> honest <=count + Gemini-voiced report cards; graceful fallback to the white-box + templated cards - LLM via opencode.ai/zen gateway (llm.py): Opus chat + direct-OpenAI embeddings - Run LLM work off the event loop (asyncio.to_thread) so the Redis response publishes - C1: dedicated Postgres (app/db/) persists the per-user feed; get_scout_feed replays it so matches survive navigation/refresh - match contract + report-card fields (schema.py); skills.py; tests/	2026-06-19 15:51:16 +05:30

Author

SHA1

Message

Date

raulgupta

be921559e7

Real Scout dashboard backend + per-board cursor rework

- get_scout_stats aggregator (app/engine/stats): assembles the REAL Summary metrics —
  funnel + cohort engagement rank + active-window (activity timestamps by hour) from our DB,
  accumulating salary band (avg of each deck's peak, ₹L), match/competition stats from the
  feed, Momentum/QX + Q-Score trend (qscore-service), day streak (user-service). Honest:
  unsourced cards return None so the UI omits/locks them, never faked. posted_date extractor.
- Activity tracking: viewed/saved flags + search_count → funnel (Matches→Viewed→Shortlisted→
  Applied) + engagement percentile. matchesFound = all-time count.
- Per-board search cursors {board: page} (replaces the single cursor): only boards that truly
  paginate (LinkedIn) get one; cursor = LAST page fetched (1st search of a new query → 1).
  Resets on query change OR >24h (boards refresh ~daily). Dropped Naukri incremental/stateKey
  (opaque, exhausting, cross-account dedup state) — dedup is the PER-USER seen-net only.
- tests: stats helpers (posting-age, histogram, active-window, engagement score).

2026-06-21 12:12:28 +05:30

raulgupta

b1e9cdd182

Engine v2: embeddings sift + Opus curator + Postgres feed persistence

- A1: fire the hard filter (Apify is precise); embedding vibe-ranker (embed.py)
  blends with the floor-free white-box (rank.py) to sift ~90 -> top 18
- A2: Opus curator (curate.py) reads the 18 -> honest <=count + Gemini-voiced
  report cards; graceful fallback to the white-box + templated cards
- LLM via opencode.ai/zen gateway (llm.py): Opus chat + direct-OpenAI embeddings
- Run LLM work off the event loop (asyncio.to_thread) so the Redis response publishes
- C1: dedicated Postgres (app/db/) persists the per-user feed; get_scout_feed
  replays it so matches survive navigation/refresh
- match contract + report-card fields (schema.py); skills.py; tests/

2026-06-19 15:51:16 +05:30

2 Commits