Microsoft Hackathon 2021
AI-Powered Poultry Production System
Hilights
I led the research and design in this hackathon project, which transformed manual food inspection into an AI-powered system. The project won 1st place at the Microsoft Hackathon 2021 partner group and achieved an 83% reduction in production harm after implementation.
Project Timeline:
Hackathon: Oct 2021 (72-hour sprint)
Implementation: Oct 2021 – Apr 2022 (6 months)
Role: UX Designer & Presenter
Team: 1 PM, 5 Devs, 1 Partnership Manager
Tools: Figma, Power BI
tl;dr
In Summary, here's what I accomplished:
Interviewed 3 stakeholder groups: Factory Executives, Production Line Supervisors, and the Supply Chain Team.
Designed dashboard information architecture (IA) by mapping 3 stakeholder groups’ information needs
Created the QinSight dashboard to visualize AI recognition data for tracking factory safety and supplier quality.
Presented the prototype to Microsoft judges, winning 1st place among partners
After implementation
Reduced production harm by 83%
Saved 264 labor hours monthly
In 2021
Microsoft Azure plans to expand its market into the manufacturing industry.
WiAdvance
- Microsoft Gold Partner & Wistron (Fortune 500) subsidiary
- Strong manufacturing networks in Taiwan
- Plan to implement Azure services in real manufacturing scenarios
Qin - Chicken Supplier
- Qin, Taiwan’s major chicken supplier
- Needed to reduce costs and increase efficiency through digital transformation
- Health-conscious consumers demanded better food safety transparency
Research & Discovery
Understanding Quality Control Challenges
Key Stakeholder Insights
Through interviews with COO and on-site observations, we identified two critical needs:
- Managers needed data-backed decisions for supplier quality control
- Workers needed real-time feedback on chicken quality without disrupting their workflow
How Might We
Transform manual quality inspection into an AI-powered system that enables real-time, data-driven decisions?
Business Requirements (Qin)
- Real-time quality monitoring system
- Traceable data for farm supplier evaluation
- Cost-effective solution to replace manual inspection
Technical Requirements (Microsoft)
- Implement Azure AI
- Use Power BI for data visualization
- Showcase Microsoft’s manufacturing capabilities
Where to
Start ?
Strategic Moves: Where AI Adds Most Value
Process Mapping
Mapped the quality inspection journey:
- Workers manually inspect chickens for defects
- Managers record data in spreadsheets
- Quality reports created at day’s end
- No real-time data for quick decisions
Solution 1
Real-time Quality Monitoring with AI
Record Defect Data with Azure AI
Trained Azure Vision AI to identify 4 defect types, each telling a different story:
- Empty hooks: Worker efficiency issues
- Wings defect: Rough handling during hanging
- Head defect:
- Bruised breast: Poor farm treatment
- Head defect: Improper blade positioning
Ensure Process Safety with AI
- Broke down chicken handling into 4 key movements
- Mapped hand washing into 7 hygiene steps
- 🚨 System alerts focus on safety, not worker surveillance
- Warning sounds prevent harmful movements before injury
Fun Fact
In Taiwan, workers and managers use LINE (like WhatsApp) for everything – even factory alerts.
Solution 2
Data-Driven Farm Evaluation
Key Metrics
We use artificial intelligence to collect data in the production line, including the actual number of chickens, the defect rate, and the average weight. Actual numbers differing from supplier data, high rates of breast bruising defects, and average weights that are too light are all evidence that a farm is not up to standard.
Award
Microsoft Hackathon 2021 Partner Group - 1st place
Being a gold partner of Microsoft in Taiwan, WiAdvance was privileged to receive an invitation to Microsoft’s 2021 Hackathon, focusing on AI industry applications.
I proudly presented our project, the Azure AI food traceability system, which was in active development. The result was truly remarkable as we won first place in the Hackathon, underlining our commitment to excellence in AI and technology.
Outcomes
From Data to Action
After implementing the AI quality control system:
- Reduced production harm by 83% through real-time movement monitoring
- Saved 264 labor hours monthly by automating quality inspection.
- Won 1st place in Microsoft Hackathon 2021 Partner Group
Key Learnings
- Start small but impactful – focusing on 4 key defect types proved more valuable than trying to solve everything at once
- Balance automation with human expertise – AI should enhance, not replace, worker knowledge
- Real-time feedback is crucial – immediate alerts prevent issues before they become problems
An Intel AI-powered mental health companion rapidly designed to provide safe, inclusive support for LGBTQ+ individuals.
A hardware design leveraging data visualization to transform testing insights into ergonomic monitor solutions.