Aphex

AI Developer

Philippines, PhilippinesRemoteFull timePosted 18 days ago
Apply on Aphex →

Sign into see who you know at Aphex.

ABOUT APHEX

We're the construction execution platform replacing outdated spreadsheets with multiplayer tools that delivery teams love. Major contractors like Acciona, CPB Contractors, John Holland, and McConnell Dowell use Aphex to deliver projects from £50 million to multi-billion pound mega projects.

We're market-leading in Australia and the UK, supporting teams building the roads, rail, tunnels, and hospitals that millions use every day. We're leaning further into AI as a core part of where the product goes next, and we've started building the foundational pieces.

Construction has been underserved by technology for decades, and it's now at real risk of being left behind by AI. We don't intend to let that happen. Aphex builds the orchestration layer the world's major construction projects run on, and AI is central to where we take it next. This role is for someone who wants to be at the heart of that, not just shipping AI features, but helping shape how an entire industry puts AI to work.

You'll start hands-on, building and shipping our AI features in the real product. From there it grows. As you learn what works, you'll help set the standard for how we build, evaluate and monitor AI across Aphex. Practical first, then broadening into the bigger picture.

WHAT YOU'LL DO

BUILD AND IMPLEMENT

- Translate briefs into working AI features, selecting the right model and implementation approach for the use case

- Test inputs and outputs before shipping

- Ship AI features into the real product

EVALUATION AND RELIABILITY

- Monitor and evaluate AI feature performance in production

- Dig into real inputs and outputs to find where prompts or implementations can be better

- Keep every AI feature across the platform accurate and reliable over time

- Build the evaluation coverage that catches regressions before users do

IMPROVE AND ITERATE

- Refine prompts and implementations based on how features actually behave in the wild

- Research evaluation...