Contamination model on trucks
Hypothetical. Not a client, not a result.
// The problem
A recycling program loses loads to contamination, but cannot tell which routes or addresses are causing it, so education goes everywhere and changes nothing.
// What we would build
A custom vision model, fine-tuned on the program's own hopper camera images, that flags likely contamination as each cart is tipped, tied to the cart's RFID lift record, so the program can send targeted notices to the addresses that need them.
// What is in it
- Hopper camera images captured at each lift
- Likely contamination flagged by type, for a reviewer to check
- Flags tied to the address through the cart's RFID lift record
- Targeted notices and follow-up for repeat addresses
- A model fine-tuned on your images and measured on a held-out set
// Stack
- PyTorch
- ONNX Runtime
- Python
- Next.js
- AWS GPU instances
// estimate
- Build
- ≈ US$168,000 to US$257,000, delivered within 41 weeksCAD 239,900 to 366,300
- Hosting
- ≈ US$7,020 a monthCAD 10,000 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 4.4 weeks
- Core features
- 1.9 to 3.4 weeks
- Full build
- 1.4 to 2.4 weeks
- Training
- 8.9 to 15.4 weeks
- Testing and fixes
- 0.4 weeks
- Evaluation and red-teaming
- 3.9 to 4.4 weeks
- Report checked against a past period
- 3.9 to 4.4 weeks
- Launch
- 0.9 to 1.4 weeks
- Production
- 0.9 to 1.4 weeks
Outside our hands, and added to the calendar
- Truck readers and cart tags delivery
- 2 to 6 weeks
- Fleet GPS API access from your provider
- 1 to 4 weeks
How it is paid
- Deposit 20%
- CAD 47,740 to 73,020
- Discovery 1.1%
- CAD 2,625.70 to 4,016.10
- Data and evaluation plan 7.2%
- CAD 17,186.40 to 26,287.20
- Design approved 1.1%
- CAD 2,625.70 to 4,016.10
- Core features 3.5%
- CAD 8,354.50 to 12,778.50
- Full build 2.3%
- CAD 5,490.10 to 8,397.30
- Training 21.7%
- CAD 51,797.90 to 79,226.70
- Testing and fixes 1.1%
- CAD 2,625.70 to 4,016.10
- Evaluation and red-teaming 14.4%
- CAD 34,372.80 to 52,574.40
- Report checked against a past period 8.7%
- CAD 20,766.90 to 31,763.70
- Launch 1.1%
- CAD 2,625.70 to 4,016.10
- Production 7.8%
- CAD 18,618.60 to 28,477.80
- Holdback, 30 days after launch (10%)
- CAD 23,870 to 36,510
- GPU time for training and testing, at cost
- CAD 1,200. Billed up front, at cost, outside the milestones