Senior Full Stack Engineer

Hydra X is a MAS-licensed digital capital-markets infrastructure platform — tokenisation, institutional custody, bilateral dealing, and the AGX digital exchange. We are in the middle of a multi-phase transformation to rebuild the way we design, build, and operate software — with AI agents as the default execution layer across the SDLC, and guardrails suitable for a regulated environment.

We are hiring a Senior Full-Stack Engineer who is backend+devops heavy and works AI-native by default . You will own the build of production systems that Hydra X runs on — services that integrate with third-party APIs, enforce strict data and audit boundaries, and ship with the kind of test coverage you would want to show a regulator. Some of these systems will be used daily by Ops, Compliance, and Risk; some will sit closer to our clients and our regulated offerings. You will work on both.

You will typically:

• Own and maintain Go-based HTTP services backed by PostgreSQL.
• Build the thin but polished Next.js dashboards and web surfaces that the people using the system actually use.
• Wire up conversation-based workflows where that is the best interface for the job.
• Take a system from spec to staging UAT to production, including CI/CD and deployment onto our shared AWS platform.

You are expected to do most of this by orchestrating AI coding agents (local / background) across the SDLC, rather than typing production code.

Our team has pushed our engineering specs to a point where a capable engineer, using Cursor (or equivalent) well, can ship a full system (backend frontend tests deployment) in weeks rather than quarters. We need another engineer in the seat to keep the throughput up without dropping our compliance and security bar.

The constraint is no longer typing speed. The constraint is judgement — which part of a build goes to a background agent, which part needs a tight local loop, which failure modes must have named tests before an agent writes a line, and which architectural decisions must stay with a human because the model will otherwise pick the locally-elegant-but-globally-wrong option.

We want someone who can hold that line while still shipping fast.

This is not a “bring Copilot as an autocomplete” role. It is also not a “you must already run 6 parallel background agents” role. We are looking for someone in the middle — clearly moving in the AI-native direction, comfortable being coached on our specific patterns, and allergic to going back to the old way.

Concretely, by the end of your first 60 days, we expect you to be able to:

• Use Cursor (or equivalent) as your primary IDE , with local agents handling a meaningful share of day-to-day work — not just autocomplete.
• Hand off self-contained components to background agents from a spec we provide, converge them via a test-driven loop, and return a mergeable pull request.
• Drive QA through agents — generate test matrices from specs, close coverage gaps, generate realistic fixtures, identify weak tests, and run integration suites against real dependencies (containerised PostgreSQL, mocked external services). You treat tests as the holdout set that defines “done,” not as a post-hoc checkbox.
• Recognise the common failure modes of AI code generation — hallucinated APIs, plausible-but-wrong library versions, tautological tests, silent mocking of the thing under test, spec/implementation drift, secrets in diffs — and build small habits around catching each before merge.
• Write specs that agents can execute without you hand-holding every step. You will co-author specs with the team lead and, over time, start authoring your own.

If you already work this way, great — you will fit quickly and probably push the team further. If you have strong fundamentals and are moving in this direction but not fully there yet, that is also fine — we will pair you with our most AI-fluent engineers during onboarding and expect you to be fully productive by month two.

What we are not hiring for:

• Someone who resists agent-driven development on principle.
• Someone who treats AI tooling as an optional add-on.
• Someone who wants to hand-write everything because it feels more “real.”

That working style is incompatible with how the team ships today.

The systems we build tend to share a shape, regardless of whether they face Ops, Compliance, Risk, engineering, or our clients:

• A Go HTTP service (Gin or similar) that integrates with one or more third-party APIs, enforces business rules, writes to a PostgreSQL database designed for auditability, and exposes a versioned REST API documented in OpenAPI.
• A Next.js web surfac