Senior Software Engineer (Optimisation & Decision Science)

We are a team of leaders, bold thinkers and humble learners and we're here to overcome big challenges, build solutions for people to enjoy, while doing the best work of our careers, together, as Bash. We use technology and talent to bring the physical and digital worlds closer, and we're transforming SA retail for the better.

Bash is one of South Africa's leading omnichannel retailers with the transaction volumes, customer and product data, and operational complexity to make the problems here genuinely hard. The optimisation opportunity is significant and largely untapped, and we're expanding the team to pursue it properly.

THE ROLE

The Optimisation & Decision Science team builds the models and algorithms that drive many of Bash's most important operational decisions. This is a small, rigorous team applying serious mathematics to hard operational problems. 

We need a Senior Software Engineer to join the team and work alongside our operations research engineers to take that work into production. You’ll be responsible for translating models into reliable, observable systems running at the scale the business needs. You'll bring consistency to how we develop software and deploy models, and you'll build the platforms and standalone applications that let us put automated decisions in front of real customers and operators. You'll set the engineering standard on the team and you'll own what you ship.

WHAT THE TEAM DOES

Our work sits across the business, from supply chain to customer experience to finance. Typical problems we solve:

• Planning Service is Bash’s real-time supply chain “brain” that turns orders and live fulfilment events into versioned delivery plans (SLAs + predicted dates + allocations) that unlock accurate promise dates and better customer outcomes.
• Production decision systems. Real-time services that make operational calls automatically. Fraud detection, recommendations, customer churn, propensity scoring, RFM segmentation.
• Optimisation engines. Standalone applications like Equilibrium, our stock transfer  optimiser, which has already driven millions in incremental sales and margin.
• Forecasting. Revenue, parcel volumes, demand, capacity. These forecasts feed directly into how the business plans and executes.
• Model deployment platform. The shared standards, tooling and infrastructure that make it easy to take a model from notebook to production without reinventing the wheel each time.
• Partnerships across business units. We work with fulfilment, logistics, marketing, merchandising and finance to put automated decisions where they create value.

WHO YOU ARE

• You care deeply about code quality, testability and maintainability, and you raise the standard of the engineers around you.
• You think about latency, reliability, observability and rollback from the start, not as an afterthought once something is "done".
• When you see five teams deploying models five different ways, your instinct is to design the one approach that works for everyone.
• You're good at working with data scientists and operations research engineers. You can take their prototypes and turn them into systems that run unattended in production without losing the intent of the original model.
• You're at home in AWS and you understand the cost, security and operational trade-offs of the choices you make.
• You're happy to get stuck into CI/CD, infrastructure as code and monitoring when it helps the team ship reliably.
• You take ownership of what you ship, including when it wakes someone up at 2am.
• You write clear documentation and runbooks, because you've been on the receiving end of code that doesn't have any.

WHAT YOU HAVE

Core Requirements

• 5+ years of software engineering experience with significant time spent shipping production systems.
• BSc in Computer Science, Engineering or equivalent. Strong work experience can substitute.
• Expert-level Python. You write idiomatic, well-tested, performant Python and you have a clear view on how to structure larger Python services.
• Hands-on experience deploying and operating optimisation and/or ML models in production. You've dealt with model drift, data skew, retraining cadence and silent failure modes.
• AWS experience spanning some combination of ECS or EKS, Lambda, SageMaker, Step Functions, S3 and RDS or DynamoDB.
• DevOps fundamentals: CI/CD, infrastructure as code (Terraform or similar), containerisation (Docker).

Technical Skills

• Production model serving. Batch, real-time API, streaming. You understand the trade-offs between them.
• Software engineering fundamentals. API design, testing strategy (unit, integration, contract), versioning, dependency management.
• Standalone applications. You can build full applications. Data layer, API, scheduling, operational tooling, the whole thing.
• Databases. Solid SQL, plus comfort with both relational (Postgres) and analytical (Snowflake) stores.
• Monitoring and observability. Logging, metrics and tracing in production using tools like CloudWatch or Prometheus and Grafana.
• Performance engineering. You can profile a service, find the bottleneck and fix it when there's a latency budget to meet.

Platform & Operations

• CI/CD pipelines for services and models.
• Comfortable with on-call, incident response, post-mortems and SLO-driven thinking.
• Awareness of data security and privacy, particularly around customer and transactional data.

Collaboration & Communication

• You've worked closely with data scientists or operations research engineers before and it's gone well.
• You can set technical standards across a team through code review, RFCs and pairing.
• You've mentored other engineers and raised the level of the people around you.

Advantageous

• Go (Golang). Production experience, especially on performance-sensitive or concurrency-heavy components.
• Graph databases. Memgraph, Neo4j or similar, for network and relationship problems.
• Temporal or similar workflow engines. Durable execution for long-running decision processes.
• Optimisation libraries and solvers: OR-Tools, Gurobi, CPLEX, Pyomo.
• Workflow orchestration: Airflow, Step Functions, Prefect.
• Stream processing: Kafka, Kinesis.
• Background in retail, ecommerce or supply chain.
• MLOps tooling (MLflow, Weights & Biases) and feature stores.
• Experience with experimentation platforms and A/B testing infrastructure.

WHAT SUCCESS LOOKS LIKE

In your first 6 to 12 months:

• We have a standard, opinionated way of deploying models across ODS, and you designed it.
• Several of our existing ad hoc deployments have been moved onto that standard.
• You've shipped at least one new production decision system yourself, something like a fraud, recommendation, forecasting, churn, RFM or propensity service.
• The engineering quality of the team is noticeably higher, through code review, mentoring and the tooling you've put in place.

We ask a few optional demographic questions to monitor fairness in hiring and comply with South Africa’s Employment Equity requirements. Your responses are confidential - stored separately from your application and never shown to hiring panels. We review the data only in aggregate to strengthen our sourcing so pipelines are diverse and all candidates receive equitable support. You can skip any question.