diff --git a/internal/playlists/seed_selection_test.go b/internal/playlists/seed_selection_test.go
new file mode 100644
index 00000000..6dc28136
--- /dev/null
+++ b/internal/playlists/seed_selection_test.go
@@ -0,0 +1,47 @@
+package playlists
+
+import (
+	"testing"
+
+	"github.com/jackc/pgx/v5/pgtype"
+)
+
+// pickSeedArtistsFromRows is the pure helper that turns sqlc rows into
+// the seed list. The DB-level query is exercised in system_test.go's
+// integration test; this unit test pins the post-fetch logic.
+
+func TestPickSeedArtistsFromRows_PreservesOrder(t *testing.T) {
+	mk := func(b byte, score int64) seedArtistRow {
+		return seedArtistRow{
+			ArtistID: pgtype.UUID{Bytes: [16]byte{b}, Valid: true},
+			Score:    score,
+		}
+	}
+	// Caller (SQL) already orders + limits; helper preserves order.
+	rows := []seedArtistRow{mk(1, 100), mk(2, 50), mk(3, 25)}
+	got := pickSeedArtistsFromRows(rows)
+	if len(got) != 3 {
+		t.Fatalf("len: got %d, want 3", len(got))
+	}
+	if got[0].Bytes[0] != 1 || got[1].Bytes[0] != 2 || got[2].Bytes[0] != 3 {
+		t.Errorf("expected [1,2,3] preserving input order; got %v", got)
+	}
+}
+
+func TestPickSeedArtistsFromRows_FewerThanThree(t *testing.T) {
+	mk := func(b byte, score int64) seedArtistRow {
+		return seedArtistRow{ArtistID: pgtype.UUID{Bytes: [16]byte{b}, Valid: true}, Score: score}
+	}
+	rows := []seedArtistRow{mk(1, 100)}
+	got := pickSeedArtistsFromRows(rows)
+	if len(got) != 1 {
+		t.Errorf("len: got %d, want 1", len(got))
+	}
+}
+
+func TestPickSeedArtistsFromRows_Empty(t *testing.T) {
+	got := pickSeedArtistsFromRows(nil)
+	if len(got) != 0 {
+		t.Errorf("empty input should yield empty output; got %v", got)
+	}
+}
diff --git a/internal/playlists/system.go b/internal/playlists/system.go
new file mode 100644
index 00000000..93549218
--- /dev/null
+++ b/internal/playlists/system.go
@@ -0,0 +1,67 @@
+// Package playlists' system.go implements the system-generated mix
+// builder (M7 #352 slice 2). The cron loop and lazy fallback both call
+// BuildSystemPlaylists; the helpers in this file (pickSeedArtists,
+// pickRepresentativeTrack, tieBreakHash) are the pure parts.
+package playlists
+
+import (
+	"hash/fnv"
+	"sort"
+
+	"github.com/jackc/pgx/v5/pgtype"
+)
+
+// seedArtistRow mirrors the sqlc-generated PickSeedArtistsRow shape.
+// Defined locally so unit tests don't need a real DB.
+type seedArtistRow struct {
+	ArtistID pgtype.UUID
+	Score    int64
+}
+
+// pickSeedArtistsFromRows projects sqlc rows into the seed list. The
+// SQL already orders by score DESC + artist_id and LIMIT 3, so this is
+// just a column projection — but pulling it into a function keeps the
+// call-site readable and makes the post-fetch path testable without
+// a database.
+func pickSeedArtistsFromRows(rows []seedArtistRow) []pgtype.UUID {
+	out := make([]pgtype.UUID, 0, len(rows))
+	for _, r := range rows {
+		out = append(out, r.ArtistID)
+	}
+	return out
+}
+
+// tieBreakHash returns a deterministic 64-bit hash of (track_id, date_str).
+// Used to break score ties in mix candidate ranking. Same inputs always
+// produce the same output; different inputs almost always differ.
+//
+// Uses FNV-1a 64-bit (stdlib hash/fnv) — fast, deterministic, no extra
+// dependencies. The exact hash isn't load-bearing; any deterministic
+// 64-bit hash works.
+func tieBreakHash(trackID pgtype.UUID, dateStr string) uint64 {
+	h := fnv.New64a()
+	if trackID.Valid {
+		_, _ = h.Write(trackID.Bytes[:])
+	}
+	_, _ = h.Write([]byte(dateStr))
+	return h.Sum64()
+}
+
+// rankedCandidate is a (track_id, score) pair used during in-memory
+// sorting before insert into playlist_tracks. T5 fills these from
+// recommendation.Candidate scores.
+type rankedCandidate struct {
+	TrackID pgtype.UUID
+	Score   float64
+}
+
+// stableSortByScoreThenHash sorts candidates in-place by score DESC,
+// breaking ties with tieBreakHash(track_id, dateStr).
+func stableSortByScoreThenHash(cands []rankedCandidate, dateStr string) {
+	sort.SliceStable(cands, func(i, j int) bool {
+		if cands[i].Score != cands[j].Score {
+			return cands[i].Score > cands[j].Score
+		}
+		return tieBreakHash(cands[i].TrackID, dateStr) < tieBreakHash(cands[j].TrackID, dateStr)
+	})
+}
diff --git a/internal/playlists/tiebreak_test.go b/internal/playlists/tiebreak_test.go
new file mode 100644
index 00000000..db538ec2
--- /dev/null
+++ b/internal/playlists/tiebreak_test.go
@@ -0,0 +1,33 @@
+package playlists
+
+import (
+	"testing"
+
+	"github.com/jackc/pgx/v5/pgtype"
+)
+
+func TestTieBreakHash_Deterministic(t *testing.T) {
+	id := pgtype.UUID{Bytes: [16]byte{1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16}, Valid: true}
+	a := tieBreakHash(id, "2026-05-04")
+	b := tieBreakHash(id, "2026-05-04")
+	if a != b {
+		t.Fatalf("same inputs gave different hashes: %d vs %d", a, b)
+	}
+}
+
+func TestTieBreakHash_DifferentDateChangesHash(t *testing.T) {
+	id := pgtype.UUID{Bytes: [16]byte{1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16}, Valid: true}
+	may4 := tieBreakHash(id, "2026-05-04")
+	may5 := tieBreakHash(id, "2026-05-05")
+	if may4 == may5 {
+		t.Errorf("different dates should change hash; got %d for both", may4)
+	}
+}
+
+func TestTieBreakHash_DifferentTrackChangesHash(t *testing.T) {
+	a := pgtype.UUID{Bytes: [16]byte{1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16}, Valid: true}
+	b := pgtype.UUID{Bytes: [16]byte{16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1}, Valid: true}
+	if tieBreakHash(a, "2026-05-04") == tieBreakHash(b, "2026-05-04") {
+		t.Errorf("different track ids should change hash for the same date")
+	}
+}