TEXTILE OPERATIONS × AI · OPEN TO GLOBAL CONVERSATIONS
From the knitting floor to AI-driven manufacturing.
Textile Operations Leader transitioning into AI-Driven Manufacturing. Eighteen years of running knit-fabric production — now turning shift logs, stop reasons and KPI sheets into live control rooms, early warnings and decisions a plant can act on.
An operator learning AI in public — with receipts.
I am not selling a title. I am an experienced manufacturing leader building practical AI-driven workflows on real production data — and showing the work.
What I bring
✓Eighteen years of floor-to-department leadership — I know what a stop reason, a GSM miss and a late T&A actually cost.
✓Working dashboards on real shift and KPI data, not slideware.
Buyer-specific fabric development, bulk and subcon scheduling.
05
2017–25
DGM, Fabrics Division
KPI/PMS, budgets, lean planning — ~55 MT/day to dyeing.
AI
2025 → NOW
Head of Dept. + AI-driven ops
Leading knitting while building control rooms and decision dashboards.
03 / Proof of work
Two live dashboards. Open them, click around.
Built from real production and KPI data, then anonymised for public view — no company names, no personal names. Drag the 3D floor, filter the quarters, drill into a machine.
Find the lost hours.
A 28-day shift log for a 37-machine circular knitting floor, turned into a diagnosis: which machines, which shifts, which stop reasons — and a what-if plan to recover the time.
●Privacy by design: operator and staff names and IDs are replaced with anonymous labels, and no employer is named. Figures are shown as they appear in the source data.
Inside the three systems
What each one actually does.
Three tools, one habit: turn a spreadsheet into a conversation a plant can act on. Here is what is inside each: the data it reads, how it works, the decisions it drives, and where it is honest about its limits.
BUILTLIVE ON THIS PAGE
Knitting Floor Control Room
Which machines, shifts and stop reasons are costing the most hours, and what would recovering them be worth?
Data in
28-day shift log from 37 circular knitting machines, three shifts (A 06–14, B 14–22, C 22–06)
Operator stop-reason log for September
What is inside
3D machine floor: height and colour show the chosen measure; click a machine for its diagnosis
Machine ranking, machine-by-day heatmap and shift trend
Operator league by quartile, performance matrix and machine allocation (anonymised on this page)
Stop-reason Pareto with the 80% band, and loss families grouped by who can act
Daily efficiency control chart with auto-detected alerts
What-if simulator, priority actions and a 90-day roadmap with owners and targets
How it works
Control limits are the mean ± 2.66 × average moving range, so normal day-to-day noise is not chased
A data-check panel lists what to fix at the source so the numbers stay trustworthy
Decisions it drives
Which three or four stop reasons to attack first
Who coaches whom, and what the daily production meeting watches
Honest limits
Efficiency moves with running time because machine speeds are held fixed
The simulator is a modelled estimate, not a promise
BUILTLIVE ON THIS PAGE
KPI Performance Dashboard
Where did the department lose KPI points, who needs a plan, and who can teach the rest?
Data in
FY 2024 department scorecard: 25 people, 18 KPIs, four quarters, with weights and targets
What is inside
Score by level and quarter as a percentage of target, with the Q1 to Q4 change
KRAs grouped into four balanced-scorecard perspectives, with the top shortfalls
Improvement plans with 30-60-90 day milestones, coaching for rating 3, recognition for 4 and 5
Skills matrix showing who can teach a skill and where nobody can
Board-review mode for the quarterly rating review
How it works
Person score is the sum of weight times achievement; quarters without data are skipped
Peers who already do well on the same KPI are shown as internal buddies
Decisions it drives
Who gets an improvement plan, a coach or recognition
Which KPI definitions to fix first, and where to train
Honest limits
The source file disagrees with itself in places; a Data checks page lists the contradictions, unclear items and a recalculated total
Production views by time-and-action week, shipment week and frame are not built because the file lacks those fields
BUILTPRIVATE · FINDINGS ONLY
NPT Improvement Tracker
Is non-productive time really falling since the devices went live, and where should the next fix and training go?
Data in
Stop logs from NPT devices on 37 circular knitting machines (reliable from 6 August 2026), per-machine shift logs and operator reason reports
An inputs workbook for targets, costs, the training log and the action plan
What is inside
Control room with a 3D floor and replay by week
Board review: weekly and monthly NPT hours and percentage, with a present-as-slides mode
Machine scorecard, reason map by loss family and A, B, C shift views
Operator skills matrix (improve, monitor, trainer) and a training plan with schedule and evaluation
ROI and value view with payback and sensitivity, and a 90-day improvement plan
How it works
Drop new Excel exports in a folder and double-click once to rebuild: daily, weekly or monthly
Cleaning rules: roll-out period excluded, duplicates removed, stops over 8 hours treated as extended stops
Operators are graded on the gap to what their machines would expect, not on raw NPT
What it found
NPT fell from 41.3% of loaded time in August to 39.1% in October: real, but modest, and flat since September
About a third of NPT minutes still have no reason code
Stops over 15 minutes are 40% of NPT minutes but under 2% of stop events
Decisions it drives
Reason-code discipline, alerts on long stops and a yarn-breakage task force
Coach on the gap, not on raw NPT
Honest limits
Data before 6 August is unreliable and excluded; device sensitivity changed, so minutes are compared, not events
Money values rest on placeholder yield and margin inputs, so none are shown here
Operator NPT percentage exists only for periods with an operator summary export
Not embedded: the full dashboard holds operator-level data, so only findings are shown
04 / What the data said
From a spreadsheet to five decisions.
The point of a dashboard is not the chart. It is the conversation it forces on Monday morning. Here is what these two systems surfaced.
Floor efficiency
Count efficiency · 28-day shift log
Target 65% — gap of 6.7 pts. The white tick marks the target on the ring. Shift C 60.7% · Shift A 56.3%
Where the non-productive time goes
Stop-reason Pareto · share of total NPT minutes
Unclassified (N/A)
35.1%
Yarn breakage
30.6%
Roll cutting
8.9%
Program change
6.5%
9 other reasons
18.9%
Insight: 35% of lost time has no reason code. Fix the data first — you cannot manage what is logged as “N/A”.
What recovering the time could mean
What-if simulator · per 28 days · same machines, same speeds
58.3%efficiency today
→
69.5%modelled after actions
2,577 hmachine-hours recovered
+3.8machines’ worth of capacity
+11.2 ptsvs 65% target
MODELLED ESTIMATE · NOT A GUARANTEED RESULT
KPI performance, quarter by quarter
Department score, % of KPI target · FY 2024 · 25 people · 18 KPIs
78.0Q1
76.4Q2
83.5Q3
85.6Q4
▲ +7.6 pts Q1 → Q4 · FY score 80.8%
Where the KPI points were lost
Share of total shortfall to target · top six levers (the top five make up 65%)
Grey fabric T&A
23%
Plan efficiency
12%
Overall NPT
11%
Order & yarn closing
10%
Change-over time
9%
Fabric rejection
7%
PDF
One-page case study
The Control Room story and the NPT follow-up on a single A4 page: problem, data, findings, modelled impact, device-based NPT results and the 90-day plan. Anonymised and ready to forward.
Choose the kind of conversation, answer a few quick questions, and the form prepares a structured email in your own mail app. Nothing is stored on this site.
08 / Credentials
Technical depth. Executive perspective.
Executive MBA
North South University · Management & Supply Chain · 2025
B.Sc. Textile Engineering
Bangladesh University of Textiles · Fabric Manufacturing Technology · 2006
