Computer vision inspection · Large organisation (250 staff or more) · Engineering and aerospace
Computer vision quality inspection for a large aerospace supplier
A camera-based inspection system for a large aerospace supplier that flags defects on the line, trained on its own products and trialled alongside human inspectors before anyone relies on it. Traceability and quality-system rules mean more validation and documentation.
Illustrative cost
£37,000 to £113,000 ex VAT
- Typical
- £64,000
- Duration
- About 20 weeks
- Effort
- 78 to 150 days
Illustrative estimate, not a quote. Real prices vary.
Worked out from UK freelance contractor day rates in job adverts, with no consultancy or agency margin added. Consultancies and agencies usually charge more.
Source: IT Jobs Watch. Rates read on ; costs computed on .
What is included
- Feasibility study with sample images
- Image capture set-up and labelling
- Training and validating a defect detection model
- Integration with the line or an edge device
- A side-by-side trial against manual inspection
Deliverables
- Trained model with validation results
- Edge deployment
- Trial report
- Retraining guide
Assumptions
- Cameras and lighting are bought by the client
- Access to the line for capture and trial
- A few hundred examples of each defect type
Not included
- Mechanical rejection hardware
- Camera hardware costs
- Certification of the system
Project plan
Phases run one after another, about 20 weeks in all.
| Phase | Weeks | Main tasks |
|---|---|---|
| Feasibility and capture | 3 | Define defects; Capture sample images; Feasibility check |
| Labelling and training | 7 | Label images; Train models; Validate |
| Edge integration | 5 | Deploy to edge device; Line integration; Alerts |
| Line trial | 3 | Side-by-side trial; Tune thresholds |
| Handover | 2 | Documentation; Operator training |
Who does the work
Person-days per role in each phase, low and high effort.
| Phase | Role | Days (low to high) |
|---|---|---|
| Feasibility and capture | Data scientist | 9 to 15 days |
| Feasibility and capture | Machine learning engineer | 3 to 9 days |
| Labelling and training | Machine learning engineer | 24 to 42 days |
| Labelling and training | Data scientist | 9 to 18 days |
| Edge integration | AI engineer | 12 to 24 days |
| Edge integration | Cloud engineer | 3 to 9 days |
| Line trial | Machine learning engineer | 6 to 12 days |
| Line trial | Test engineer | 6 to 12 days |
| Handover | Project manager | 6 to 9 days |
How the cost adds up
| Role | Low (days × 25th percentile) | Typical (days × median) | High (days × 90th percentile) |
|---|---|---|---|
| Data scientist | 18 × £481 = £8,658 | 25.5 × £600 = £15,300 | 33 × £788 = £26,004 |
| Machine learning engineer | 33 × £500 = £16,500 | 48 × £564 = £27,072 | 63 × £738 = £46,494 |
| AI engineer | 12 × £488 = £5,856 | 18 × £589 = £10,602 | 24 × £863 = £20,712 |
| Cloud engineer | 3 × £428 = £1,284 | 6 × £513 = £3,078 | 9 × £675 = £6,075 |
| Test engineer | 6 × £350 = £2,100 | 9 × £438 = £3,942 | 12 × £588 = £7,056 |
| Project manager | 6 × £451 = £2,706 | 7.5 × £540 = £4,050 | 9 × £714 = £6,426 |
| Total, rounded | £37,000 | £64,000 | £113,000 |
How these estimates are worked out
Effort: for each phase we estimate how many days each role works, as a low and a high case. These are our estimates for a typical project of this kind, not measured data.
Rates: each role's daily rate comes from UK contract job adverts over the last six months, as published by IT Jobs Watch. These are freelance contractor rates as advertised, not what was paid, and not what a consultancy or agency charges.
Low end: the low-effort days at each role's 25th-percentile rate. We use the 25th percentile, not the 10th, because the lowest effort at the cheapest rates gives a floor few buyers would ever meet.
Typical: the mid-point between low and high effort, at each role's median rate.
High end: the high-effort days at each role's 90th-percentile rate.
Totals are rounded to the nearest £500 below £10,000 and the nearest £1,000 above, and exclude VAT. Phases run one after another.
Not included:
- Software licences, model and hosting fees
- Travel and expenses
- Your own staff's time
- A consultancy's or agency's margin, overheads and risk
Real prices vary with the supplier, the state of your data and systems, how much of the work your own team does, where the work happens and how much risk the supplier carries. Consultancies and agencies usually charge more than these contractor rates, and a fixed price usually costs more than time and materials because the supplier takes the risk.
Sources
Source: IT Jobs Watch: UK contract job adverts quoting a daily rate, 6 months to 15 November 2026. Each role's page was read on . Published under CC BY-NC-SA 4.0.
| Role | IT Jobs Watch page | Rates quoted | 25th percentile | Median | 90th percentile |
|---|---|---|---|---|---|
| Data scientist | Data Scientist | 201 | £481 | £600 | £788 |
| Machine learning engineer | Machine Learning Engineer | 92 | £500 | £564 | £738 |
| AI engineer | Artificial Intelligence Engineer | 384 | £488 | £589 | £863 |
| Cloud engineer | Cloud Engineer | 513 | £428 | £513 | £675 |
| Test engineer | Test Engineer | 245 | £350 | £438 | £588 |
| Project manager | Project Manager | 1,226 | £451 | £540 | £714 |