Software Engineer

Position Overview

A Software Engineer who takes ownership of features end-to-end and is comfortable solving problems anywhere in the stack — including the parts where LLMs, agents and tool-use are the "code". You typically have 5 years or more of software engineering experience and have shipped production systems on the cloud. You have used AI coding agents seriously in your own work and want to build the platform that lets a regulated enterprise use them safely. You may lead a feature implementation with guidance from the Principal Engineer and Tech Lead, and you work closely with the Product Lead to understand the "why" behind what we build.

Sanlam Fintech is is a small business unit in Sanlam and we are setting up a small, senior AI Engineering pod with one mandate: make AI engineering agents a safe, governed, everyday part of how Sanlam builds software. We are a net producer of technology, not a consumer — and regulated financial services cannot simply ship source code, architectural context and customer data to a vendor's cloud. The pod builds the in-tenant agentic engineering stack that makes the productivity gain available inside our own boundary.

The pod's five focus areas:

1. Context & memory — giving models the right organisational and engineering context at the right time, with team-wide RBAC in place.

2. Agentic orchestration — multi-agent workflows that take features from story to production across our services.

3. Production incident resolution — agents that triage, trace and propose fixes for live incidents.

4. Developer enablement — reusable AI skills and team context that raise the floor for every engineer.

5. Agents for engineering teams — assisting with agile workflows, backlogs and delivery visibility 

What will you do?

• Take ownership of new capabilities across the pod's five focus areas and deliver them from concept to production inside Sanlam's AWS tenant.
• Build and operate agentic workflows: multi-agent orchestration, tool integrations (MCP), context/memory services, and the guardrails (RBAC, audit logging, encryption) that make them safe.
• Integrate agents into real engineering flows — BFF/microservice feature delivery, local runners, acceptance tests, incident triage against Datadog/CloudWatch and dead-letter queues. 
• Author reusable AI skills, prompts and team-context packages that other Sanlam engineers can pick up on day one.
• Measure what you ship: evals, latency, token cost-per-feature, and reliability of agent runs; feed results into the Codeship cockpit metrics.
• Focus on operational and code excellence during code reviews and feature development; improve the team's development, testing and operational practices. 
• Participate in on-call for the platform and handle incidents effectively — using the very tooling you build.
• Identify and pay down technical debt within your scope; grow your influence beyond individual tasks to the broader team. 

Qualification and Experience

• Relevant Degree or Diploma (BSc Computer Science, IT or equivalent experience).
•  5+ years of experience in Software Engineering, with production systems on AWS or equivalent.
• Hands-on experience building with LLM APIs (e.g. Anthropic, Bedrock, OpenAI) — tool use / function calling, structured output, prompt and context management.
• Demonstrable daily use of agentic coding tools (Claude Code, Cursor, Copilot agents or similar) and opinions on where they fail. 

Programming Languages & Coding

• We are language agnostic, but proficiency in the languages we run on is required: Java (Lambda/BFF services), TypeScript (frontends, agent tooling) and Python (agents, evals, MCP servers).
• Proficiency at reading, reviewing and understanding code — including code written by agents, where review is the primary control.
• Ability to articulate language differences and the build & execution process (typing, compilation, runtime), and to reason about what an agent can and cannot safely be allowed to do. 

AI & Agentic Systems

• Understanding of retrieval, embeddings and memory patterns; knowledge graphs or vector stores; and when each is the wrong tool.
• Experience with agent orchestration frameworks or hand-rolled agent loops; MCP (Model Context Protocol) servers and clients a strong plus.
• Building evals and observability for non-deterministic systems: tracing, replay, regression suites. 
• Awareness of prompt-injection, data-exfiltration and privilege-escalation risks in agent systems and how to mitigate them. 

Data Structures, Algorithms & Problem Solving

• Ability to articulate the pros/cons of relevant data structures for memory and execution time, and to choose the right one for the situation.
• Ability to suggest and trade off algorithms when developing and optimising solutions — and to adapt when scope changes. 

Engineering & Operational Excellence

• Understanding of scalability, robustness, operational visibility and agility in distributed systems.
•  AWS experience preferred: Lambda, ECS/containers, Kinesis, DynamoDB, S3, VPC, IAM, Cognito. All relevant cloud experience is beneficial.
• Monitoring, metrics, health, fault tolerance; APIs and auth; databases and persistence; event-driven architectures. 
• Security-by-default mindset: data residency, POPIA, least-privilege RBAC, audit trails — non negotiables in this pod, not later hardening. 

What will make you successful in this role?

• You've felt the gap yourself: you know what agents can do, and you know why a bank or insurer can't just turn them on.
• You develop the ability to lead small projects or components with minimal supervision, and start delivering through others. 
• You move fast and fix things — publish early, learn from failure, go again. Each phase of the alpha ships one demonstrable outcome.
• You contribute to improving team practices and coach engineers across Sanlam on working effectively with agents.
• You deepen your understanding of business objectives — cost-per-feature, delivery visibility, CISO sign off — and how technical work supports them. 

Build a successful career with us

We’re all about building strong, lasting relationships with our employees. We know that you have hopes for your future – your career, your personal development and of achieving great things. We pride ourselves in helping  our employees to realise their worth. Through its five business clusters – Sanlam Fintech, Sanlam Life and Savings, Sanlam Investment Group, Sanlam Allianz, Santam, as well as MiWay and the Group Office – the group provides many opportunities for growth and development.

Turnaround time

The shortlisting process will only start once the application due date has been reached. The time taken to complete this process will depend on how far you progress and the availability of managers. 

Our commitment to transformation