Engineering consultancy · Pretoria, South Africa · Working across Africa
Thaler is an engineering consultancy whose team has built and run production systems at Amazon, BMW Group, Absa, BVNK, Standard Bank and FNB. We deploy engineers forward, inside your team, to design, build and modernise payments, core platforms, utility systems and agentic AI for institutions across Africa.
About the firm
Our engineers have spent the last two decades inside the systems that move money and run operations across Africa and beyond: banking at Absa, Standard Bank and FNB, payments infrastructure at BVNK, automotive platforms at BMW Group and cloud services at Amazon. We work hands-on: architecture, code, operations and the mentoring that makes a platform stick after the engagement ends.
Across the continent, banking sectors are opening to new entrants, central banks are standing up instant payment systems and linking accounts to national digital IDs, and utilities are moving to prepaid metering. These programmes need engineers who have shipped the same kinds of systems in comparable markets, and who can transfer that capability to local teams. That is the work we do.
What we do
Every engagement is scoped around a concrete system and a measurable outcome. Our engineers work inside your team, bringing architecture, delivery and operations together rather than handing over a document and leaving.
Switch and instant-payment integration, mobile money and wallet rails, USSD and agent channels, reconciliation and settlement. Designed for idempotency, auditability and 24/7 operation from the first commit.
Discuss a payments projectAgents that triage incidents, resolve customer requests and run back-office workflows, built on your data and your tools. Every agent ships with authorisation boundaries, human approval steps and evaluation sets, measured in pilot before it is trusted in production.
Discuss an agentic AI pilotSenior engineers who embed inside your team and build in your environment: your repositories, your data, your release process. They ship weekly increments to production and stay until the capability is in your hands, rather than advising from a distance.
Discuss a forward-deployed teamMoving legacy JEE and monolith estates to Spring Boot services on AWS or Kubernetes, with CI/CD and infrastructure as code. Oracle-to-Aurora migrations that cut query latency 55% and database spend 28%.
Discuss a modernisationPrepaid vending and billing, field telemetry from Modbus and MQTT devices to the cloud, remote monitoring with full audit trails, and dashboards that operations teams actually use.
Discuss a utility platformDomain-driven target architectures, architecture decision records, engineering standards and SLO frameworks from an interim or fractional chief architect. Mentoring, code review, hiring support and golden-path templates so your engineers own the platform after we leave.
Discuss an architecture reviewSelected work
A selection of engagements delivered by members of our team over the last decade. Figures are from the projects as delivered and measured.
Banking · Pan-African markets
Built transactional-banking microservices for Absa's pan-African markets on Spring Boot with Kafka and RabbitMQ on OpenShift, inside a regulated environment. Implemented a high-throughput event processor on the LMAX Disruptor pattern with batch ETL from Hadoop and Hive into MongoDB, indexed with Lucene. p95 latency fell 58% and production incidents 35%.
Automotive · Cloud migration
Led the modernisation of legacy JEE services into Spring Boot microservices, containerised on AWS (ECR and EKS) with CI/CD and infrastructure as code. Migrated Oracle workloads to Amazon Aurora using AWS Glue, improving query p95 by 55%. Delivered Angular applications for plant operations and visibility.
Cloud · Engineering standards
Set backend technical direction for Java 17 microservices with clear domain boundaries, idempotent APIs and structured tracing. Introduced golden-path service templates and an SLO and alarm framework: p95 API latency fell from 210 ms to 140 ms and error-budget adherence rose from 92% to 98%. Led a Rails monolith migration to React micro frontends and Java services.
Applied AI · Operations
Designed and delivered a multi-agent on-call workflow on AWS Bedrock (Claude) with Amazon Kendra retrieval and Step Functions orchestration. It triages tickets, retrieves runbooks, executes safe actions behind tool-authorisation guardrails and drafts remediation for human review. Auto-drafted first responses for 68% of incidents and cut page volume 23% in pilot.
Industrial IoT · Fire safety
Ingested supervisory, flow and pressure signals over Modbus RTU and TCP, normalised them to JSON and published through MQTT to AWS (IoT Core, Lambda, DynamoDB). Scheduled remote test routines with exception alerting and full audit trails for NFPA-25 readiness, with Grafana dashboards for health and drift. Manual inspection hours fell 35%.
Telecoms and utilities · Citizen channels
Led a small team building a USSD platform through which municipal residents report service issues straight into the back office, plus an SMS ordering and notification platform for a national directory business. Earlier, owned the Android audit app and web modules of a utilities software suite and built the backend for a prepaid voucher payment system.
Perspectives · African markets
Four shifts we expect to shape technology programmes across the continent, and the experience we bring to each.
Payments
Central banks across Africa are standing up instant payment systems and moving their markets to ISO 20022. Real-time rails expose weaknesses that batch systems hide: duplicate postings, slow reconciliation and services that cannot run around the clock. The institutions that do well will treat this as an architecture decision rather than an integration ticket.
Identity
Regulators across the continent are requiring financial accounts to be linked to national digital IDs, often on tight deadlines. Most institutions do not need a new core to meet this. They need a well-governed integration layer that handles consent, verification, audit and exceptions, and that works over USSD and agent channels as well as apps.
Energy
Across the continent, utilities are moving customers to prepaid metering while revenue protection and outage visibility remain hard problems. Vending, token management, billing and field telemetry tend to live in separate systems. Bringing them onto one event-driven platform is what turns a meter rollout into an operations capability.
Agentic AI
Banks and utilities are being sold autonomous agents at every conference. The useful version is narrower, and it is built inside your operation rather than bought off a shelf: agents that triage incidents, retrieve runbooks and draft actions for human approval, with explicit tool-authorisation boundaries and evaluation sets that measure answer quality before and after launch.
How we work
The last phase makes us unnecessary. That is the point.
One to two weeks reviewing architecture, code, operations and the team. Findings are ranked by risk and payback, not by what is fashionable.
Deliverable: written findings and a prioritised list of options.
Target architecture, bounded contexts and data ownership, migration path, and a sequenced plan with costs and owners attached to every line.
Deliverable: architecture decision records and a costed roadmap.
Forward-deployed engineers build alongside yours, in your environment, in weekly increments, with CI/CD, observability and runbooks from the first release rather than bolted on later.
Deliverable: working software in production, measured against the targets we set together.
Playbooks, standards, golden-path templates and training so your team runs and extends the platform without us.
Deliverable: documented operations and a team that owns the system.
Fixed scope, two to four weeks. Findings, target state, decision records and a costed roadmap your team or any vendor can execute.
Milestone-based or time and materials. Our engineers embed in your team and build in your environment, from a payments integration to a full modernisation programme.
A retainer for senior technical direction a few days a month: design reviews, vendor evaluation, hiring and board-level reporting.
Workshops, pairing and code review that raise engineering standards and leave written practices behind.
Working across borders
We are based in Pretoria and work across African time zones. Engagements run remote-first with on-site blocks wherever your team sits.
We respond to RFPs and RFQs, and we work in consortium or subcontract arrangements with local partners where procurement rules require them. Contracting runs through a registered South African company, with company documents supplied alongside proposals.
The team
Thaler is a consultancy of senior engineers with experience from Amazon, BMW Group, Absa, BVNK, Standard Bank and FNB.
Between us we have built pan-African transactional banking services, digital banking channels, payments infrastructure, plant and booking platforms for a global carmaker, and the service templates and AI tooling that keep a cloud provider's fleet running. We have delivered in South Africa, Nigeria and Canada, on-site and remote, in regulated environments where mistakes are expensive.
Every engagement is led by a senior engineer who has shipped the kind of system you are building, and that person stays hands-on from architecture through to production. The specialists we bring in for payments, data, mobile channels, cloud or agentic AI work the same way.
Amazon · BMW Group · Absa · BVNK · Standard Bank · FNB
Java, Go, TypeScript, Kotlin, SQL · AWS (Lambda, Step Functions, EventBridge, SQS/SNS, DynamoDB, ECS/EKS, KMS, IAM) · Docker, Kubernetes, OpenShift
PostgreSQL, Oracle and Aurora, MySQL, SQL Server, MongoDB, Redis · Kafka, RabbitMQ · Hadoop/Hive, Lucene, OpenSearch, Amazon Kendra
AWS Bedrock (Claude), agent orchestration with Step Functions, retrieval-augmented design, LangGraph and LangChain patterns, guardrails and offline evaluations · Domain-driven design, event-driven architecture, SLOs and observability, Scrum and Kanban
Contact
Tell us about the system, the deadline and the constraints. We will come back with an honest read on whether and how we can help.
Phone and WhatsApp
Location
Pretoria, South Africa
Working with clients across Africa
Tenders and RFPs
Send the terms of reference
We confirm fit, partnering arrangements and availability before you commit time to a bid
We typically reply within two business days.