Talentflow

TalentFlow (talentflow.fyi) — Company-to-company talent-flow graph: a directed movement matrix built by querying public professional-profile search data for profiles whose current employer is B and whose prior work history lists A, then keeping only the aggregate result count c(A,B) per ordered pair. The published layer applies SpringRank on the movement graph plus net-movement weights w(A,B)=max(c(A,B)-c(B,A),0), a source-quality weighting S(B), a rank-realism inclusion rule and an evidence-confidence shrinkage, to produce a percentile-ranked Company Flow Index banded S/A/B/C/D/E (top 10 / n…

Talentflow offers (Alternative, Sentiment) — Stated per-company evidence counts give a real handle on volume: indexed companies carry evidence from a few hundred to six figures — Google 155,270; Amazon 169,362; Meta 73,258; Apple 84,991; Wells Fargo 105,624; Bank of America 101,289; NVIDIA 34,692 — while the top-ranked company shows 5,166. Summed across an indexed company set N, the pairwise matrix is order N^2 potential edges with observed evidence counts in the low millions of profile-matches, so the raw matrix is in the 10^5-10^7 observed edge range with per-edge counts from single digits to tens of thousands (inferred from the published per-company evidence figures and the graph's inclusion thresholds).

Competitive talent-flow and labour-market network analysis (who hires from whom, which firms are net importers — a structure not available from job postings); recruiting strategy and sourcing-map design (where candidates at a target firm previously came from and where they go next); compensation and offer benchmarking context by identifying which firms compete for the same talent pool; equity/credit research as a soft signal of a company's ability to attract talent from strong sources (a leading indicator of capability and of competitor erosion); VC/growth diligence on startups that lack financial disclosure (the top bands are dominated by recently founded AI companies); workforce-planning and org-design analysis; anti-poaching and retention risk scoring; and graph-learning training data — the movement matrix is a real-world directed weighted graph suitable for link prediction, node ranking and network-embedding benchmarks.