Unominda

The constraint

What the system had to solve before a line of it was designed.

Unominda manufactures car seats for automotive assembly lines across multiple production shifts. Every count was manual — operators filled paper sheets at shift end, supervisors consolidated them into Excel, and by the time the number reached the planning team it was hours stale and frequently wrong. Worse, quality holds were logged separately from production counts, so the 'completed' number included seats already flagged and pulled by QC. The planning team was making assembly commitments against numbers they couldn't trust, and the floor had no real-time visibility into where they stood against the shift target.

Architecture

The stack read top down, from what the user touches to what keeps it running in production.

Modules delivered

Each one shipped production-grade, not as a demo surface.

• 01Barcode scanner integration at all assembly stations
• 02Real-time production count dashboard (line, shift, daily)
• 03QC checkpoint flow — holds, reworks, and scrap tracking
• 04Shift target management with live pace alerts
• 05Automated shift handover report generation
• 06Management portal with plant-wide visibility
• 07Assembly-ready count feed for downstream planning

Delivery

4 months in total, staged so each phase reached production before the next one started.

1. 01Discovery to production deployment across 3 assembly lines

Design decisions

What we learned before building, and what the built system moved.

We spent two weeks on the shop floor across all three shifts before writing a single line of code. The research revealed that the count problem was fundamentally a trust problem — no two people in the building agreed on how many seats were truly 'done' at any moment. The planner's definition, the floor supervisor's definition, and the QC team's definition were all different, and none of them had a shared data source. We traced every physical step a car seat takes from raw frame to packed-and-labelled, mapped each point where the paper system lost track of state, and designed scanner touchpoints around those exact gaps. The QC gate logic was co-designed with the quality team over three working sessions — they defined the pass/fail rules and rework paths, we built the engine that enforces them and makes every decision traceable. We piloted on a single line for two weeks before rolling to all three — both to validate the scanner UX under real production pace and to give the floor teams time to trust the number on the screen before we decommissioned the paper sheets.

The nightly count reconciliation meeting was cancelled on day three of the live rollout — there was nothing left to reconcile. Seat count accuracy moved to 100% verified. The QC pass rate at dispatch climbed to 99.1% because defects were now caught at the line within the same shift, not discovered at the loading dock the next morning. The planning team started receiving a clean assembly-ready count four to six hours earlier in the day than before, which directly improved their next-day scheduling accuracy. The floor supervisors' shift reports, previously a 45-minute manual process at the end of every shift, now generate automatically with a single confirmation tap and reach the management portal before the supervisor has left the building.