AI/ML Engineer

# AI ML Engineer • *Squad B**

## What We're Building

Softsure is the technology company behind Dotsure, one of South Africa's leading pet insurance providers. For 15 years, our platform has processed policies, claims, payments, and integrations. It works. But it's a VB.NET monolith with a single SQL Server database, and it's reaching the limits of what that architecture can support.

We're not throwing it away. We're modernising it piece by piece using the strangler fig pattern: new capabilities in C#, Blazor, and Azure services, gradually replacing the legacy code while the system keeps running. By 2030, we'll have a modern, event-driven, API-first platform that other business units can build on.

Squad B is the team that makes this happen. These six roles form the platform engineering capability that will build the foundations: the data platform, the CI/CD pipelines, the API layer, the AI integrations, and the developer experience tooling.

## What We're Looking For

Every role requires experience migrating legacy systems to modern architectures. We don't need people who've only worked on greenfield projects. We need people who understand the messiness of real systems: the undocumented business rules, the integrations that nobody remembers building, the edge cases that only appear in production.

We need people who can look at a 15-year-old codebase and see both what it is and what it could become.

## AI/ML Engineer • *Put AI into Production, Not Just Prototypes**

We're not looking for someone to build demos. We're looking for someone who can put AI into production and keep it running. Claims document extraction. Fraud detection. Customer service copilots. These aren't science projects. They're operational capabilities that need to work reliably every day. If you've shipped AI to production and maintained it, we want to talk.

### About This Role

Softsure processes thousands of insurance claims, each with documents that need to be read, validated, and processed. Our contact centre handles customer queries that follow predictable patterns. Our fraud detection relies on rules that could be augmented with ML. We have real problems that AI can solve. We need someone to solve them responsibly.

### What You'll Do

- Define the AI governance framework: how models get approved, tested, monitored, and retired - Implement Azure OpenAI integrations with proper guardrails (content filtering, PII handling, audit logging) - Build the claims AI pipeline: OCR, document classification, entity extraction, validation - Establish prompt engineering standards and build a reusable prompt library - Create evaluation frameworks so we can measure accuracy, latency, and cost objectively - Implement responsible AI controls because insurance data is sensitive and regulators are watching - Optimise AI costs. Token budgets, caching strategies, model selection. This stuff adds up fast - Support the Contact Centre Copilot implementation - Roll out GitHub Copilot to the development team with appropriate policies and training - Monitor model performance in production and catch drift before it causes problems

### What You Bring

- You've deployed AI/ML to production and maintained it. Jupyter notebooks don't count - 5+ years in software engineering with 2+ years focused on AI/ML - Hands-on Azure OpenAI or OpenAI API experience - Strong Python skills for ML pipeline development - Experience with document AI: OCR, form recognition, NER, document classification - Understanding of LLM patterns: RAG, fine-tuning, prompt engineering, embedding models - Familiarity with vector databases and semantic search - Software engineering discipline: testing, code review, documentation, monitoring

### Education

- Degree or Diploma in Computer Science, Information Technology, Software Engineering, Computer Engineering, or a related technical field. - Relevant cloud, API, or platform engineering certifications (Azure preferred) will be advantageous. - Equivalent practical experience building and maintaining production APIs and distributed systems will also be considered.

### Nice to Have

- Azure AI Services experience (Document Intelligence, Cognitive Services) - Insurance claims processing knowledge - Experience with AI safety and responsible AI practices - LangChain, Semantic Kernel, or similar frameworks - GitHub Copilot enterprise deployment experience

### Why Join Us

- Real problems to solve, not AI for AI's sake - Budget for Azure OpenAI and the tools to do this properly - Leadership that understands AI needs governance, not just enthusiasm - Direct impact on operational efficiency (claims processing, customer service) - Competitive salary and the chance to define AI practices for the organisation

### Location & Work Arrangement

This is an in-office role based in George, South Africa. We believe the collaboration and mentorship required for a transformation of this scale happens best when the team is together.