Data Collection & Scoping

Sky

Set-top box and broadband equipment refurbishment for the UK and European markets, operated from Bydgoszcz, Poland. This page provides the complete data collection playbook for deploying the multi-modal AI optimization loop on the Sky program.

1 SiteSet-Top BoxesBroadband EquipmentHigh VolumeEU Operations

Facility Location

🇵🇱

Bydgoszcz

Kuyavian-Pomeranian Voivodeship, Poland

European operations hub for Sky set-top box and broadband equipment processing. Handles high-volume intake, functional testing, firmware updates, cosmetic grading, refurbishment, and redistribution for the UK and European markets.

Timezone: CET (UTC+1)
Complete Checklist

Data Collection Checklist

Complete every item before the AI model can begin processing. The Sky program has additional firmware-related data requirements compared to other programs.

Spatial / 3D Capture

7 items
1
Schedule Matterport Pro3 scan of entire Bydgoszcz facility
2
Capture: receiving/intake area, triage/sorting, functional test stations, firmware update stations, cosmetic grading, refurbishment benches, packaging, and shipping/palletization
3
Scan material flow paths between areas — especially conveyor routes and buffer zones
4
Capture storage areas (incoming inventory, WIP buffers, finished goods staging)
5
Export as Matterport Digital Twin (.e57 point cloud + web link)
6
Minimum resolution: 4K (3840x2160) per scan position
7
Estimated time: 5–7 hours (Sky facilities tend to be larger due to high-volume throughput)

Video Capture

10 items
1
Record 15–30 minute continuous video at EACH workstation type (triage, test, firmware, cosmetic, refurb, packaging)
2
Capture at least 2 full shifts to show operator variation and shift-to-shift differences
3
Resolution: minimum 1080p (1920x1080), preferred 4K. Frame rate: 30fps minimum
4
Landscape orientation ONLY — mount iPad/iPhone on tripod at operator eye level
5
Include wide-angle establishing shots showing the full line layout (1 per area)
6
Capture close-up detail shots: set-top box disassembly, PCB inspection, firmware flashing, cosmetic assessment
7
Record dock-to-stock flow: follow a pallet of Sky devices from receiving through to shipping
8
Include audio — capture ambient sound, conveyor noise, test equipment beeps/alerts
9
Label each video file: SKY_BYD_{AREA}_{STATION}_{DATE}_{SHIFT}.mp4
10
Capture the triage/sorting process in detail — this is where the AI can have the highest initial impact

Plus (MES) / IFS (ERP) Data

11 items
1
Extract 30 days of production data from Plus covering the video capture period + 30 days prior
2
Include: production orders, cycle times per station, defect/failure codes, yield rates, throughput per SKU variant
3
Extract BST (Bill of Standard Times) data for every station captured on video
4
Pull operator performance data (anonymized by operator ID) for the same period
5
Extract defect/failure code distributions: which fault codes are most common for Sky devices?
6
Export material flow data: units received, units processed, units shipped per day per SKU
7
Pull IFS data: cost per unit, material costs, labor costs, CAPEX/OPEX for Sky program
8
Extract firmware version tracking data: which firmware versions are being flashed, success/failure rates
9
Generate data dictionary listing every field, type, unit, and meaning
10
Export in JSON (API) + CSV (bulk) formats
11
Configure MCP connector for real-time Plus data access from the Bydgoszcz instance

Audio Capture

6 items
1
Deploy audio recording devices at each major work area
2
Record minimum 2 hours continuous per area during peak production
3
Format: WAV (uncompressed), 44.1kHz sample rate, 16-bit depth minimum
4
Capture machine sounds: test equipment cycling, conveyor belts, packaging machinery
5
Capture ambient noise levels at different times of day
6
Label each file: SKY_BYD_{AREA}_{DATE}_{TIME_START}-{TIME_END}.wav

Equipment & Labor Data

8 items
1
Complete equipment registry: every machine, test station, and tool used in Sky processing
2
Include: equipment ID (matching Plus), make/model, age, PM schedule, last PM date, condition rating
3
Export PM completion logs and downtime logs with reason codes for the past 6 months
4
Compile labor roster: operator ID, role, contract type, shift assignment, hire date, certifications
5
Export skill/certification matrix: which operators are qualified for which stations and SKU types
6
Pull CAPEX records for equipment purchases in the past 24 months
7
Export OPEX data: maintenance costs, consumables, tooling replacement costs per station
8
Note: Polish labor regulations may affect data collection — verify with local HR

Enrichment Data

8 items
1
Collect BOMs for every Sky SKU variant processed (set-top boxes, routers, broadband equipment)
2
Gather technical spec sheets for each device model (component specs, test parameters, firmware requirements)
3
Compile all work instructions (WIs) for each operation
4
Collect quality standards and cosmetic grading criteria (Sky-specific requirements)
5
Gather firmware update procedures and version compatibility matrices
6
Include Sky customer requirements documents and SLAs
7
Collect any existing process flow diagrams or value stream maps
8
Include any existing training materials for new operators

Folder Structure

Copy this exact folder structure. Note the additional firmware_procedures/ directory specific to the Sky program.

📂 Sky/
  📂 Bydgoszcz/
    📂 01_Spatial/
      📄 matterport_scan_{DATE}.e57
      📄 matterport_web_link.txt
      📄 floor_plan_annotated.pdf
    📂 02_Video/
      📂 receiving_intake/
      📂 triage_sorting/
      📂 functional_test/
      📂 firmware_update/
      📂 cosmetic_grading/
      📂 refurbishment/
      📂 packaging_shipping/
      📂 establishing_shots/
      📄 video_index.xlsx
    📂 03_Plus_IFS_Data/
      📂 raw_exports/
      📂 aggregated/
      📄 data_dictionary.xlsx
      📄 mcp_config.json
    📂 04_Audio/
      📄 SKY_BYD_{AREA}_{DATE}_{TIME}.wav
    📂 05_Equipment_Labor/
      📂 equipment/
      📂 labor/
      📂 financial/
    📂 06_Enrichment/
      📂 boms/
      📂 spec_sheets/
      📂 work_instructions/
      📂 quality_standards/
      📂 firmware_procedures/
    📄 collection_checklist.xlsx
    📄 README.md

7-Day Collection Sprint

The Dream Team travels to Bydgoszcz and completes all data collection within 7 business days. Note: coordinate with Polish public holidays and local labor regulations.

Day 1
Matterport 3D scan of entire facility (5–7 hours)
Set up video capture equipment at all stations
Introductory meeting with site management and shift leads
Day 2
Video capture: Shift 1 — all stations (triage, test, firmware, cosmetic, refurb, packaging)
Audio recording deployment across all areas
Day 3
Video capture: Shift 2 — all stations (variation capture)
Dock-to-stock flow recording (follow pallet end-to-end)
Close-up detail captures: PCB inspection, firmware flashing, cosmetic assessment
Days 3–5
Plus (MES) data extraction via API/MCP from Bydgoszcz instance
IFS (ERP) data extraction
Equipment registry and labor roster compilation (coordinate with local HR)
Days 5–7
Enrichment data collection (BOMs, spec sheets, WIs, firmware procedures, quality standards)
Data validation and cross-referencing against Plus records
Upload to shared folder structure and complete checklist
Exit briefing with site management

AI Optimization Targets

Triage/sorting optimization — reduce decision time and improve routing accuracy for incoming devices
Functional test station utilization and throughput optimization per SKU variant
Firmware update success rate improvement — reduce retry rates and identify root causes of failures
Cosmetic grading consistency across operators — reduce subjective variation
Line balancing across the full refurbishment workflow
Buffer zone optimization — minimize WIP inventory between operations
Conveyor routing efficiency — optimize physical flow paths based on SKU mix
Operator skill-to-station matching for maximum throughput
Predictive maintenance for test equipment based on usage patterns and failure rates
Shift-to-shift performance variance reduction
Packaging and palletization efficiency for UK/EU distribution
Energy consumption optimization (relevant for EU sustainability reporting)

Dream Team Sky Assignments

What each Dream Team member specifically does for the Sky data collection sprint in Bydgoszcz.

AI Model Coach

Define optimization targets specific to set-top box and broadband equipment workflows
Prioritize triage/sorting as the highest-impact area for initial AI optimization
Validate that video captures show representative production conditions across SKU variants
Review initial AI outputs and provide structured feedback on routing and line balancing recommendations
Define firmware update success rate thresholds — what constitutes an anomaly worth investigating?

Data Capture Specialist

Operate Matterport Pro3 for 3D scan of the Bydgoszcz facility (5–7 hours)
Set up and manage video capture equipment at each station type
Extract Plus data via API from the Bydgoszcz instance — ensure firmware tracking data is included
Configure MCP connector for real-time Plus data access
Deploy audio recording devices and manage file naming conventions (SKY_BYD_ prefix)
Coordinate with local IT for network access and data export permissions

Operations Analyst

Coordinate with Bydgoszcz site manager to schedule data collection during representative production days
Compile equipment registry and labor roster (coordinate with local HR — note Polish labor regulations)
Map the current process flow — identify Sky-specific operations (firmware update, triage/sorting)
Track collection progress against the checklist and flag gaps before the team leaves
Document shift patterns and labor scheduling practices specific to the Polish facility

Engineering & Quality Lead

Collect BOMs for every Sky SKU variant (set-top boxes, routers, broadband equipment)
Gather firmware update procedures and version compatibility matrices — this is unique to Sky
Validate that test parameters and cosmetic grading criteria are documented per Sky's requirements
Ensure defect code definitions in Plus are consistent and complete for Sky devices
Collect Sky customer SLAs and quality requirements documents

Project Manager

Plan the Bydgoszcz trip: travel, site access, equipment shipping, local contacts
Coordinate with Polish public holiday calendar to avoid scheduling conflicts
Run daily standups during the sprint; produce end-of-sprint completion report
Manage the master checklist — ensure all 6 categories are complete before departure
Coordinate with Bydgoszcz site management for exit briefing and next-steps alignment

Sky-Specific Considerations

Firmware Version Complexity

Sky devices require specific firmware versions based on model, region, and customer tier. The AI model needs firmware version tracking data to optimize the update process and reduce retry rates.

High-Volume Throughput

Sky's Bydgoszcz facility processes significantly higher volumes than typical repair programs. The AI model must account for conveyor-based flow and batch processing patterns.

EU Data Regulations (GDPR)

Ensure no personally identifiable information is captured in video or audio recordings. Operator data must be anonymized. Coordinate with local DPO if needed.

Cosmetic Grading Subjectivity

Sky has specific cosmetic standards that vary by market (UK vs. EU). The AI model can help standardize grading decisions and reduce operator-to-operator variation.

Seasonal Volume Variation

Sky volumes fluctuate seasonally (higher around holidays and product launches). Data collection should capture a representative period, not a peak or trough.

Polish Labor Regulations

Shift patterns, break requirements, and overtime rules differ from US/Mexico facilities. The AI model must account for these constraints in optimization recommendations.