Open Study College

The short version

Ohana was already around four years old when I joined Open Study College. My work has focused on helping it grow into a faster, cheaper, serverless and event-driven platform—and making it easier for the engineering team to keep moving it forward.

I was promoted to Senior Software Engineer within six months and went on to lead technical initiatives across application code, infrastructure, authentication, real-time systems, developer tooling and product rollout.

Modernising Ohana

The aim was not to rewrite a working product for the sake of it. We moved parts of the platform towards serverless and event-driven patterns where they simplified the system or made it less expensive to operate.

That work reduced cloud costs by roughly 90% for the migrated workloads. It also made the platform easier to change without tying every new feature to long-running servers.

Making the important paths faster

Production measurements guided the performance work:

• p50 latency fell by 70%
• p95 latency fell by 30%
• key request paths went from roughly three seconds to under 50ms

The numbers belong to specific paths and workloads rather than being a claim that every request became instant. The useful part was finding the work that users felt most and fixing that first.

Go services and tooling

I built an authentication service in Go that made authentication around 300ms faster per request.

I also built a Go-based language server to help migrate our internal Tailwind-like design-token system to Tailwind CSS. It brought migration guidance into engineers’ editors, making a broad frontend change easier to adopt incrementally and consistently.

Those were part of a wider set of internal tools built with Go to remove repetitive work and make the development experience better.

Shipping without losing an afternoon

Deployments that had taken hours were reduced to under two minutes. Faster feedback changed more than the stopwatch: smaller changes became easier to release, verify and roll back.

I also helped introduce controlled feature rollout and PostHog experimentation, so product changes could be measured rather than released on instinct alone.

AI-assisted engineering practices followed the same principle. The goal was useful leverage with engineers still responsible for the plan, review and outcome—not outsourcing judgement to autocomplete.

Real-time chat

Chat was part of Ohana too. Reworking its real-time architecture saved roughly £5,000 per year while keeping the experience responsive for learners and staff.

Work beyond Ohana

Open Study College has more than one product. I helped reduce hosting for its older applications from roughly £20,000 to £1,000 per month, without attributing that saving to Ohana.

I also built reusable Sanity CMS foundations for the ecommerce site, making content-driven features easier to extend without rebuilding the same integration each time.

What I took from it

The best modernisation work is rarely the flashiest. It is usually a series of well-measured changes that make the product faster, the bill smaller and the next engineer’s job easier.