Snapwire

One-Third of Their Engineering Team

Snapwire wasn’t a stock photo library. It was an on-demand photography marketplace. Dell needs product photos at their Austin office? Starbucks needs lifestyle shots at a specific location? A brand needs 200 images on a topic by Friday? Snapwire matched them with professional photographers, handled the logistics, processed the deliverables, and split the payments.

Fortune 500 clients. Tens of thousands of photographers. Millions of images through the pipeline.

We provided 10 of the 30 engineers for 2.5 years. Not a rescue mission — Snapwire’s CTO ran a well-organized operation out of Toronto. The internal team was split between Canada and California. They had quarterly planning, specced sprints, and a codebase that was already several years into production. They needed capacity. We were the capacity.

Our tech lead came in with 15 years of experience. He’d previously managed 90 servers on call at NOSphere Ventures. The team under him covered backend, frontend, mobile, QA, DevOps, and project management. Same codebase. Same sprints. Same code review process as the Toronto and California engineers.

The Hard Part Was Images

The entire platform revolved around media. That sounds obvious, but the implications aren’t.

Photographers upload raw files from dozens of different camera manufacturers. Each upload needs to be processed into multiple formats — cropped previews for browsing, scaled versions for client review, and full-resolution originals for final delivery. All stored on S3, delivered through CloudFront CDN. The previews need to appear instantly while the full-resolution version is still processing. And you can’t lose color fidelity or metadata during any of the transformations.

We built the cropping tools, the CDN storage pipelines, and the content management interface where brands browsed, reviewed, selected, and downloaded their commissioned photos. Getting images from camera to client without losing quality or adding latency — that was the engineering challenge, not “how do you store files on AWS.”

Matching, Search, and Scoring

The matching engine connected brands with the right photographers using multiple signals: availability (calendar integration), past experience with similar shoots, subject matter tags, quality scores from completed assignments, and geography for on-site work. No point matching a New York photographer for a shoot in San Diego.

AWS ML services handled automated image tagging — identifying objects, scenes, and themes — plus quality scoring for sharpness, exposure, and composition. These fed back into the matching algorithm and supplemented manual curation.

Elasticsearch powered search and discovery on both sides. Brands searched for photographers by specialty, location, and portfolio. Photographers searched for available assignments. The index updated in near-real-time as new registrations, assignments, and portfolio images came in.

Marketplace Plumbing

Stripe Connect handled payment splitting. Brand pays Snapwire. Snapwire takes commission. Photographer gets their payout. Per-assignment pricing, bulk project pricing, and usage-based licensing — all through the same infrastructure.

We rebuilt the frontend and designed registration wizards for both sides. Multi-step flows with validation at each stage, designed to minimize drop-off. Photographer registration: portfolio, availability, payment setup, specialty tags, equipment. Brand registration: company verification, brief templates, budget config, team access.

Pre-moderation kept quality high. Photographer portfolios reviewed before going live. Assignment deliverables reviewed before release. Automated flagging with manual review. The platform served Fortune 500 clients and the moderation reflected that.

Stack and Numbers

Laravel. React. PostgreSQL. Elasticsearch. AWS (S3, CloudFront, ML services). Stripe Connect. 10 of our engineers in a 30-person team. 2.5 years continuous. Millions of images. Dell, Starbucks, Fortune 500s.

Acquired by StudioNow. The engineering passed due diligence — which is what happens when the codebase is clean enough that strangers can read it, evaluate it, and plan to extend it.

What This Shows

When a 30-person team needs to move faster, you either hire internally (3-6 months to find, interview, onboard, ramp) or you add a vetted external team that integrates from week one. We were the second option.

The engineers on Snapwire worked in the same sprints, reviewed each other’s code, and shipped to the same standards as the internal team. That’s not “outsourcing.” That’s team augmentation done properly.