AI Software Engineer

Everyone's racing to build with AI. Far fewer people can make it hold up in production: handle real traffic, fail gracefully, stay observable at 12:34am, and cost what it's supposed to cost. That gap, between a model that demos well and a service customers can actually depend on, is where this role lives. At Transcom, our AI/ML Engineers and Researchers push the models forward.

You build everything around them that turns a clever prototype into software people trust: the APIs, the pipelines, the fallbacks, the telemetry. The craft that decides whether an AI feature ships or stalls. If you're a strong backend engineer who's genuinely curious about AI but would rather own the system than tune the model, this is the seat for you.

You'll own the engineering: how AI capabilities get exposed, integrated, deployed, and kept alive, on an event-driven platform that processes customer interactions at scale. You'll work closely with the AI/ML team, but the software craft is yours. In this role you will: Design and build production-grade backend services and APIs that expose AI/ML capabilities: LLMs, SLMs, STT, TTS, NMT, embeddings, classification - to our products and external systems.

Take research prototypes from the AI/ML team and turn them into scalable, secure, reliable software that survives real traffic. Integrate AI services with robust error handling, monitoring, graceful degradation, and clear fallback paths for when a model is slow, down, or wrong. Design and optimise data and model-serving pipelines for performance, reliability, and cost.

Own deployment: CI/CD, environment management, and the automation. Make AI systems observable - logging, tracing, metrics, dashboards, and alerts on latency, reliability, cost, and quality. Treat evaluation as first-class: build the harnesses, quality gates, and regression tests that catch hallucinations, latency regressions, and quality drift before customers do.

Keep it safe and compliant: secure handling of PII, access control, encryption, GDPR-aligned practice. Make the team compound - code reviews, technical talks, and mentoring that turn individual learning into team leverage.