1. MLOps & Azure Architecture
You’ll architect and build our entire cloud-native MLOps platform on Azure, including:
- High-throughput video processing infrastructure
- Robust data pipelines (Encord → Azure → Roboflow)
- Automated model training + versioning
- Deployment of high-performance inference services
- ETL pipelines to turn predictions into client-ready metrics
This is full system ownership — and you’ll be the one building it.
2. Computer Vision Strategy & Model Lifecycle
You will define how we track brand assets across sports video and social media content by:
- Designing the CV approach (models, data strategy, optimisation)
- Integrating Encord annotations with Roboflow training pipelines
- Deploying and optimising the model at scale
- Setting up monitoring to catch model drift before it becomes a problem
3. Human-in-the-Loop (HITL) Tooling
You’ll build the analyst verification interface that:
- Flags low-confidence detections
- Lets analysts quickly confirm/correct outputs
- Feeds improved data straight back into the training pipeline
This is the backbone of our continuous learning loop.
4. Knowledge Transfer
At the end of the project you’ll hand over:
- Documentation
- Architecture diagrams
- Runbooks
- Codebase walkthroughs
So our internal team can take full ownership.
🌟 Why This Role Is Special
- True ownership: You design it, you build it, you ship it.
- No legacy tech: Clean slate. You choose the right tools and architecture.
- Real-world impact: Your system will power visibility metrics used by global brands.
- Massive scope: MLOps, CV, cloud architecture, inference, ETL, HITL — all yours.
- Fast-moving domain: Sports + AI + video = one of the most exciting fields right now.
If you’re tired of being one engineer on a huge team and want to own the entire machine, this is your chance.