OIA AI
Background
At Thought&Function, we know that startups with big ideas often face a tough reality—especially when technical expertise becomes a barrier to realising their vision. Our mission is to bridge that gap, offering tailored, cloud-native solutions that make advanced AI and data products not just possible, but scalable and impactful.
That’s exactly the kind of partnership we built with Oxford Immune Algorithmics (OIA). OIA, a health tech innovator based in Oxford, was creating groundbreaking AI models to help physicians diagnose blood-related conditions faster and more accurately. But like many growing startups, they were running into challenges:
- Outdated infrastructure that couldn’t scale with their ambitions.
- Limited in-house expertise in data engineering and cloud optimization.
- A need for faster processing and easier maintenance without ballooning costs.
They came to us for help, and together, we transformed their machine learning data pipeline into a high-performance, future-ready system.
Challenge
OIA’s flagship product, Algocyte, uses Artificial General Intelligence (AGI) to analyse blood samples and deliver insights at the point of care. It’s the kind of tool that could save lives. But their existing infrastructure was holding them back:
- Their machine learning models were powerful but slow, hampered by inefficiencies in their current setup.
- Adding new AI models or scaling up required significant time and effort.
- Managing the infrastructure was complex and resource-intensive.
They needed a solution that could handle not only their current AI models but also scale seamlessly as they added more features and functionality.
Team
What we did
When we work with clients, we don’t just deliver solutions—we build them together. For OIA, this started with deep collaboration. We sat down with their team to fully understand their needs, their vision, and the technical challenges they were facing. Then, we got to work.
Here’s how we approached it:
- Evaluating the Current Pipeline
We began by reviewing OIA’s pipeline and identifying pain points. This wasn’t just about spotting inefficiencies; it was about creating a roadmap for improvement that aligned with their business goals. - Designing a Future-Ready Infrastructure
We proposed several solutions, carefully balancing trade-offs between cost, performance, and upkeep. Together, we chose a cloud-native design on Google Cloud Platform (GCP) that prioritised scalability and minimised maintenance.- Tools of the Trade: We used GCP’s Cloud Functions, Bigtable, Pub/Sub, and Compute Engine to build a distributed system capable of processing large data volumes while dynamically scaling to meet peak demands.
- Building an End-to-End AI Pipeline
The new pipeline automatically picked up blood sample images, processed them through multiple steps, and ran them through nearly a dozen AI models in parallel. This made it possible to detect analytes like platelets and white blood cells quickly and accurately.- Smart Scaling: The system could instantly handle higher workloads by spinning up new cloud functions when needed—without wasting resources during downtime.
- Seamless Integration: We created an API that made results instantly accessible for OIA’s app, ensuring the diagnostics could be delivered where they mattered most.
Ensuring Smooth Deployment
To keep things simple and sustainable for OIA’s team, we automated everything. Using GitHub Actions for continuous integration and Terraform for infrastructure as code, we made sure the pipeline could be deployed and updated with ease.
The Project
The outcome
The impact was immediate and tangible:
- Twice as Fast: The new pipeline processed data in half the time of their old system.
- Effortless Scalability: It dynamically adjusted to varying workloads, making it cost-efficient and future-proof.
- Simplified Management: OIA’s team could now focus on innovation instead of wrestling with infrastructure.
- Ready for Growth: Adding new AI models or functionality became a straightforward process, supporting OIA’s ambitious plans for the future.