Automate 8M Product Image Population
Budget / Salary₹37,500–75,000
TypeFreelance project
LocationRemote
Posted1 hour ago
We have approximately **8 million products already uploaded to our website/database**, and all products currently do not have images.
We are looking for an experienced developer/automation engineer to build a **scalable automated product image sourcing, generation, matching and upload system**.
The preferred image hierarchy is:
1. Find and use an accurate manufacturer/official product image where available.
2. If a suitable real image cannot be found, use AI-generated product-only images where appropriate.
3. Products with uncertain image matches should be flagged for review rather than automatically uploading an incorrect image.
The system should use available product information such as product name, part number/SKU, manufacturer, brand, category and description to identify the correct image.
### Required Workflow
The system should automatically:
- Read products from our database/API.
- Find or generate an appropriate product image.
- Match and validate the image against the correct product/SKU/part number.
- Detect duplicate or potentially incorrect images.
- Optimize/compress images for web delivery.
- Upload images to our server/cloud storage/CDN.
- Update the correct product record with the image URL.
- Skip products that have already been successfully processed.
- Maintain processing status and logs.
- Retry failed processing automatically.
- Flag uncertain matches for manual review.
### Scale
The complete database contains approximately **8 million products**.
We do NOT expect all 8 million products to be processed immediately.
The first phase should be a **Proof of Concept (POC)** using a smaller dataset. After validating image accuracy, matching accuracy, processing speed, infrastructure requirements and cost, the system should be capable of scaling to millions of products.
### Important
This is **NOT a manual data-entry or manual image-upload project**.
We need an automated and scalable technical solution.
### When Applying, Please Answer
1. How would you architect this system for 8 million products?
2. How would you verify that an image belongs to the correct product/part number?
3. How would you source manufacturer/official product images?
4. When would you use AI-generated images?
5. How would you detect duplicate or incorrect images?
6. What technology stack would you use?
7. How would you handle failures, retries and uncertain matches?
8. What cloud storage/CDN architecture would you recommend?
9. What would be the estimated processing cost per 100,000 or 1 million products?
10. Have you previously built a large-scale image/data-processing system?
Please avoid generic proposals. Explain the architecture you would recommend specifically for this project.
If the POC is successful, this can become a significantly larger long-term project.
We are looking for an experienced developer/automation engineer to build a **scalable automated product image sourcing, generation, matching and upload system**.
The preferred image hierarchy is:
1. Find and use an accurate manufacturer/official product image where available.
2. If a suitable real image cannot be found, use AI-generated product-only images where appropriate.
3. Products with uncertain image matches should be flagged for review rather than automatically uploading an incorrect image.
The system should use available product information such as product name, part number/SKU, manufacturer, brand, category and description to identify the correct image.
### Required Workflow
The system should automatically:
- Read products from our database/API.
- Find or generate an appropriate product image.
- Match and validate the image against the correct product/SKU/part number.
- Detect duplicate or potentially incorrect images.
- Optimize/compress images for web delivery.
- Upload images to our server/cloud storage/CDN.
- Update the correct product record with the image URL.
- Skip products that have already been successfully processed.
- Maintain processing status and logs.
- Retry failed processing automatically.
- Flag uncertain matches for manual review.
### Scale
The complete database contains approximately **8 million products**.
We do NOT expect all 8 million products to be processed immediately.
The first phase should be a **Proof of Concept (POC)** using a smaller dataset. After validating image accuracy, matching accuracy, processing speed, infrastructure requirements and cost, the system should be capable of scaling to millions of products.
### Important
This is **NOT a manual data-entry or manual image-upload project**.
We need an automated and scalable technical solution.
### When Applying, Please Answer
1. How would you architect this system for 8 million products?
2. How would you verify that an image belongs to the correct product/part number?
3. How would you source manufacturer/official product images?
4. When would you use AI-generated images?
5. How would you detect duplicate or incorrect images?
6. What technology stack would you use?
7. How would you handle failures, retries and uncertain matches?
8. What cloud storage/CDN architecture would you recommend?
9. What would be the estimated processing cost per 100,000 or 1 million products?
10. Have you previously built a large-scale image/data-processing system?
Please avoid generic proposals. Explain the architecture you would recommend specifically for this project.
If the POC is successful, this can become a significantly larger long-term project.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.