Grading at processor intake
Hypothetical. Not a client, not a result.
// The problem
A processor sorts species and grades fish by eye at the intake line, so grades vary by shift and disputes with harvesters over grade are hard to settle.
// What we would build
A custom model fine-tuned on photos from your own intake line that suggests species and grade for each fish or tote, with the grader confirming or correcting each one, and a web screen that records grades against each delivery. The model learns from the corrections, and every change is checked against a held-back set of your photos before it goes live.
// What is in it
- Photos from your intake line labelled with your graders
- A model fine-tuned on your species and grades
- Suggestions the grader confirms or corrects, one tap each
- Grades recorded against each delivery and harvester
- An evaluation on a held-back set of your photos before every change
// Stack
- PyTorch
- Fine-tuned vision model
- Next.js
- PostgreSQL
- AWS GPU
// estimate
- Build
- ≈ US$122,000 to US$186,000, delivered within 31 weeksCAD 173,900 to 265,400
- Hosting
- ≈ US$6,710 a monthCAD 9,550 a month
- Support
- ≈ US$2,500 a monthCAD 3,565 a month
Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.
- AI route
- Custom LLM
- Model running cost
- ≈ US$6,530 a monthCAD 9,300 a month
Timeline by milestone
- Discovery
- 2.9 to 3.4 weeks
- Data and evaluation plan
- 0.4 weeks
- Design approved
- 2.4 to 3.9 weeks
- Core features
- 1.9 to 2.9 weeks
- Full build
- 1.4 to 1.9 weeks
- Training
- 6.4 to 10.9 weeks
- Testing and fixes
- 0.4 weeks
- Evaluation and red-teaming
- 3.9 to 4.4 weeks
- Launch
- 0.9 to 1.4 weeks
- Production
- 0.9 to 1.4 weeks
How it is paid
- Deposit 20%
- CAD 34,540 to 52,840
- Discovery 1.5%
- CAD 2,590.50 to 3,963.00
- Data and evaluation plan 7.9%
- CAD 13,643.30 to 20,871.80
- Design approved 1.5%
- CAD 2,590.50 to 3,963.00
- Core features 4.6%
- CAD 7,944.20 to 12,153.20
- Full build 3.1%
- CAD 5,353.70 to 8,190.20
- Training 23.9%
- CAD 41,275.30 to 63,143.80
- Testing and fixes 1.5%
- CAD 2,590.50 to 3,963.00
- Evaluation and red-teaming 15.9%
- CAD 27,459.30 to 42,007.80
- Launch 1.5%
- CAD 2,590.50 to 3,963.00
- Production 8.6%
- CAD 14,852.20 to 22,721.20
- Holdback, 30 days after launch (10%)
- CAD 17,270 to 26,420
- GPU time for training and testing, at cost
- CAD 1,200. Billed up front, at cost, outside the milestones