R&D Workforce · Weekly KPI

Week of 2026-09-14 · Charles Liu (Kbot-Link + Meet3) · Romit Bhingradiya (Kommerce-1 / AutoOMS)

Delta band · beat < 0.9 · on plan 0.9–1.2 · retro > 1.2. Actual ÷ required hours; three-way agreement signed before each task.

Range Scope Charles Romit

This cycle — 5 KPI band the compact numbers-view (Linear-style KPI header)

Cycle 38 · Sep 15 – Sep 21

2 engineers · 8 tasks (incl. 2 T0 access) · 43h actual / 69.5h planned

Hourly rate (avg)

$81/hr

Charles $85 · Romit $78

Hours consumed

43h / 69.5h

62% of week's plan

Total cost this cycle

$3,501

60% of $5,845 planned

Task performance

86%

↑ above 85% target

Quota met

61%

mid-cycle · on-plan trend

Charles Liu · temp assist

Kbot-Link (OTA + dashboard) · Meet3 (recording · notes · bridge) · 2 access grants active

Delta (week)

0.93

on plan

Tasks done

2 / 4

1 in progress · 1 to do

Hours

21 / 34

62% of week's plan

Blockers

clear

Romit Bhingradiya

Kommerce-1 · AutoOMS-Master · 2 repos active · English-only AI summaries

Delta (week)

1.11

on plan

Tasks done

2 / 4

1 in progress · 1 blocked

Hours

22 / 37.5

59% of week's plan

Blockers

AO-77 · needs Charles

Weekly delta trend · last 6 weeks

Delta score by week target band 0.9–1.2 shaded

Both engineers trending inside the on-plan band. Charles moved from mild retro (W33 1.15) into beat territory at W36 (0.92) and back into on-plan. Romit is holding steady around 1.1.

Table view

Week	Charles	Romit
W33	1.15	1.22
W34	1.08	1.18
W35	0.97	1.09
W36	0.92	1.14
W37	0.95	1.08
W38 · this	0.93	1.11

Task status this week stacked, share of tasks

Both at 50% task completion. Romit has one blocker (AO-77 order-status webhook, waiting on Charles's contract review) — surface for daily standup.

Required vs actual hours · per task (this week)

Charles · Kbot-Link + Meet3 21h delivered / 34h planned

Table view

Task	Required	Actual	Δ	Status
T0 · access grant	3.0	3.0	1.00	Done
T1 · OTA C+D (rollout)	14.0	13.0	0.93	Done
T2 · silent-decline consent	8.0	5.0	0.63	In progress
T3 · long-call hallucination	9.0	0.0	—	To do
Total	34.0	21.0	0.93	62% delivered

Romit · Kommerce-1 / AutoOMS 22h delivered / 37.5h planned

Table view

Task	Required	Actual	Δ	Status
T0 · access grant	3.5	3.5	1.00	Done
T1 · RMA edge cases	14.0	15.5	1.11	Done
T2 · inventory recon	11.0	3.0	0.27	In progress
T3 · order webhook	9.0	0.0	—	Blocked (Charles)
Total	37.5	22.0	1.11	59% delivered

5-KPI framework · sub-scores this week

Charles

Highest sub-score: Quality (92) — TDD pattern on OTA rollback shipped with regression tests. Lowest: Velocity (78) — one big task carried the week.

Romit

Highest: Quality (87) — real-postgres tests per CLAUDE.md Rule 3. Lowest: Focus (78) — split across two repos (Kommerce-1 + AutoOMS) reduces context depth.

Task board Linear-style grouped rows · priority · labels · assignee · delta

In progress1 8h est · 5h actual

M3-88

Recording consent — silent-decline capture compliance backend frontend

8h / 5h

△ 0.63

Sep 22

In progress

K1-204

Kommerce-1 ↔ AutoOMS inventory reconciliation infra backend

11h / 3h

△ 0.27

Sep 22

In progress

Blocked1 9h est · 0h actual · waiting

AO-77

Order-status webhook to AutoOMS backend contract

9h / —

blocked

Sep 26

Blocked

To do1 9h est · —

M3-89

Long-call summary hallucination fix backend ai

9h / —

—

Sep 26

To do

Done4 28h est · 26.5h actual · Δ 0.95

KL-142

OTA phased-rollout picker otafrontend

4h / 4h

△ 1.00

Sep 14

Done

KL-143

One-click rollback button ota

4h / 4h

△ 1.00

Sep 15

Done

KL-144

Empty-state onboarding + sample release ota

4h / 3.5h

△ 0.88

Sep 16

Done

AO-73

RMA state machine — return-label edge cases rmabackend

14h / 15.5h

△ 1.11

Sep 18

Done

R&D Workforce · Weekly KPI · Week of 2026-09-14. Data source: diary_rollups table on rnd-workforce.kasercorp.com (schema shipped in PR #1 of stevehkaser/rnd-workforce, systemd timer rnd-diary-rollup-week). Palette validated per dataviz spec — CVD-safe categorical + reserved status tones + on-plan band shading — light and dark modes both selected, not auto-inverted. Each chart has a table view fallback for accessibility.