APPLIED AI ENGINEER

WHY THIS ROLE EXISTS

Helm builds intelligent multilingual conversational assistants for major African enterprise brands. The Applied AI team is the intelligence layer behind these assistants, building and maintaining services that leverage natural language understanding, intelligent document processing, retrieval-augmented generation, and all things generative AI.

We need someone who can build production AI services and keep them running well. That means writing solid code, designing systems that hold up under real-world conditions, implementing safety and guardrails, and being rigorous about understanding when and why things go wrong. You will own services end-to-end and make architectural decisions that directly affect how enterprise clients experience our AI. You will have real ownership from early on. This is not a role where you just execute tickets, you make decisions about how services are designed, built, and operated.

WHAT YOU WILL DO

Build and operate production GenAI services. Implement LLM-powered services from architecture through to production, prompt management, provider abstraction, structured output, cost controls. The Head of Applied AI works with you to set architectural direction. You own implementation, deployment, and evolution of these services over time. As models and techniques advance, you stay close to what's changing and make grounded calls on what to adopt.

Improve retrieval systems. Work on retrieval-augmented generation, improving retrieval pipelines, response quality, diagnostics, and integration with the broader platform. Make these systems observable and systematically improvable.

Implement responsible AI practices. Content filtering, prompt injection protection, PII handling, output validation, guardrails. These are engineering problems, not checkboxes. You build them as first-class concerns in every service.

Extend observability for GenAI. Traditional monitoring doesn't cut it for LLM-based systems. You will build telemetry that tracks token costs, latency, output quality, and agent behaviour, so the team knows how systems are performing, not just whether they're up.

Drive error analysis. When something goes wrong, we need to understand why, not just patch it. You will build and contribute to systematic error analysis workflows that turn failures into actionable improvements. This is how we get better, and it's non-negotiable.

Strengthen shared infrastructure . Logging, configuration management, deployment automation, error handling. You improve the tooling and patterns that the whole team relies on.

WHAT SUCCESS LOOKS LIKE

First 3 Months

You understand the platform, the integration patterns, and the constraints. You've shipped real improvements to existing services, not just read the code. You've formed a view on where the biggest engineering leverage is and started acting on it.

3–9 Months

You've built and shipped at least one GenAI service into production with multi-tenancy, safety layer, and cost controls. Retrieval systems have better diagnostics and clearer integration patterns. Responsible AI checks are in place. You've improved shared infrastructure in ways that benefit the whole team, not just your own work.

9–12 Months

You own one or more production services end-to-end. GenAI observability is operational, the team can see what's happening, what it costs, and where it breaks. You're contributing to architectural decisions and influencing technical direction. You can evaluate new GenAI capabilities critically and recommend what's worth pursuing.

Beyond 12 Months

You're contributing to the evolution toward agentic workflows - bounded AI systems operating under clear constraints. You're helping shape how AI-assisted orchestration fits into the platform. You're mentoring new team members and your influence on system design extends across the function. The trajectory is toward senior engineering ownership, though that's about growing scope and impact, not a guaranteed title change on a schedule.

REQUIREMENTS

Required

Strong Python skills, with experience building production APIs and services (not primarily notebooks).

Experience deploying containerised services on cloud infrastructure.

Ability to reason about system architecture, API design, and production trade-offs.

Practical experience with GenAI and LLM integration, prompt engineering, RAG, or model serving. You don't need to be a data scientist, but you need to understand how these systems work and how they fail.

Comfortable working autonomously with high ownership in a small team.

Genuine interest in evaluation, error analysis, and responsible AI, not just feature shipping.

Valuable

Experience with LLM provider APIs and patterns for abstraction, cost management, and reliability.

Experience building observability and monitoring for production services.

Familiarity with FastAPI, async Python, or similar frameworks.

Experience in multi-tenant environments serving diverse clients.

TEAM AND ENGINEERING CULTURE

We value measurable impact over activity, ownership over participation, and honest assessment over optimistic reporting. We build for production, not for demos. Engineers own real systems and are expected to think critically about trade-offs. We come from diverse technical backgrounds and respect what each person brings, but we hold each other to a high standard on engineering quality.

Learning is part of how we work. We protect time for disciplined research, short cycles with clear questions and defined exit criteria. Curiosity matters, but so does accountability.