Dirk “Dirco” du Toit

Software Development Engineer · Technical Lead

linkedin.com/in/dircodt

Executive Summary

Seasoned software engineer with broad experience designing and delivering solutions across frontend, backend and data layers in scientific, engineering and commercial domains. Strongest at turning complex requirements into practical software — from architecture and prototyping through production implementation, deployment and long-term support.

Core Strengths

Software Architecture Cross-Platform Development Data Processing & Analysis Proof-of-Concept Prototyping 3D Visualisation GPGPU Computing

Selected Experience

Reutech Mining / Radar Systems — led HMI and frontend evolution across a radar portfolio, from desktop applications to a modern web-based platform with GPU-accelerated 3D/4D visualisation and data processing.

Commercial SaaS & Web Applications — designed and developed full-stack systems spanning web interfaces, backend services, databases and GIS on self-managed infrastructure.

Engineering & Scientific Software — built specialised tools across telemetry, signal processing, and interactive spatiotemporal visualisations, bridging domain expertise with software engineering.

Python · TypeScript / JavaScript · C# / .NET Data · Web · 3D · GIS
Selected Projects
Reutech Mining — Stellenbosch, South Africa circa 2017–2026
Role: Senior Software Developer | Technical Lead: Web & 3D
Stack: TypeScript · Angular · Angular Material · Three.js · WebGL · GLSL · SCSS · Turf.js · amCharts · Python · FastAPI · WebSockets · REST · MQTT · NumPy · Pillow · PostgreSQL · Docker

Modern web platform (“MSR Connect” / “MSR Solo”) for slope-stability monitoring radar, replacing a legacy desktop application with a client/server implementation that consolidates multiple radar systems into a unified operational environment with interactive 3D terrain, scan visualisations, alarms and analytical trends.

Key Contributions
  • Led the frontend and 3D architectural direction for the transition from a legacy desktop HMI to a high-performance, cross-platform client/server web platform.
  • Architected a multi-pass WebGL/GLSL shader pipeline that projects radar-scan data directly onto 3D digital terrain models (DTMs) to bypass the memory and performance bottlenecks of legacy geometry-only approaches and unlock simultaneous multi-radar visualisation in a seamless 360° environment.
  • Engineered custom GLSL shader stages covering coordinate transformation, dynamic colour ramps, Kernel Density Estimation (KDE) smoothing, real-time line-of-sight/shadow calculations and contour/grid overlays.
  • Built interactive 3D spatial tools for local georeferencing and monitoring-region vector drawing, tightly synchronising viewport interactions with hierarchical tree views, time-series charts and temporal playback controls.
  • Implemented Web Workers and WebSocket streaming to move CPU-heavy processing off the main thread and stream live radar-scan imagery while keeping the browser UI fluid and responsive.
  • Integrated third-party geotechnical sensors (such as prisms and extensometers) into unified 3D spatial views and correlated trend charts, enabling direct comparison of multi-source displacement data.
  • Evaluated performance and algorithmic trade-offs through Python/Jupyter prototypes, conducting GPU-oriented, bulk NumPy and columnar-data (Parquet/Arrow) experiments to de-risk production graphics and data pipelines.
  • Implemented full internationalisation support, including right-to-left (RTL) locales.
Outcome: Transformed a desktop-bound product into a high-performance web platform, delivering multi-radar data consolidation, GPU-accelerated terrain visualisation and analytical workflows across desktop workstations, tablets and mobile devices via any modern browser.
Reutech Mining — Stellenbosch, South Africa circa 2020–2021
Role: Senior Software Developer
Stack: Python · NumPy · SciPy · PyMeshLab · Open3D · Electron · JavaScript · CSS · Three.js · WebGL · GLSL · PyInstaller · Nuitka · Paramiko · DXF · PLY · OBJ · glTF

Cross-platform desktop application (“Deployment Helper”) designed and developed from scratch to streamline 3D terrain preparation, coordinate transformation and real-time radar coverage simulation supporting all MSR models.

Key Contributions
  • Evolved the application from a dedicated 3D mesh converter into a comprehensive interactive deployment and coverage-planning tool.
  • Built reliable mesh processing pipelines (PyMeshLab, Open3D) to preview, repair, decimate and convert DTMs, scans and point clouds—solving persistent interoperability gaps where existing third-party tools produced incompatible exports, lacked full format support or could not simplify texture-mapped meshes.
  • Implemented coordinate scaling, georeferencing and axis transformations to reconcile disparate mine survey grids, CAD conventions and radar-local coordinate systems.
  • Developed a two-pass WebGL/GLSL GPU shader pipeline to compute line-of-sight terrain visibility and surface incidence angles, enabling real-time coverage simulation across candidate radar positions.
Outcome: Unified fragmented 3D mesh preparation and interactive coverage visualisation into a single standalone tool, enabling engineers to reliably plan and optimise radar placements.
Reutech Mining — Stellenbosch, South Africa circa 2015–2016
Role: Software Developer | HMI Lead
Stack: C# · .NET Framework · WPF · XAML · MATLAB · TCP/IP · Python · NumPy · SciPy · scikit-image · OpenCV · TensorFlow.js · WebGL · GLSL

Field and desktop software for a compact FMCW ground-penetrating radar, combining real-time data acquisition with subsurface signal processing, interactive 2D radargram visualisation and geotechnical analysis.

Key Contributions
  • Independently developed the production HMI for rugged ARM tablets (field acquisition) and desktop systems (detailed analysis).
  • Built interactive 2D radargram visualisations with custom display controls, feature tracking and diagnostic analysis workflows.
  • Integrated a comprehensive signal-processing pipeline covering real-time filtering, time-to-distance conversion, FFT processing, adaptive contrast enhancement and super-resolution.
  • Ported the complete numerical pipeline from MATLAB into Python (NumPy/SciPy/OpenCV) to eliminate proprietary runtime licensing and simplify deployment.
  • Prototyped browser-based FFT and GPGPU processing using TensorFlow.js and custom WebGL/GLSL shaders to explore lightweight cross-platform architectures.
Outcome: Delivered an end-to-end radar acquisition and analysis product deployed to field tablets and desktop environments, while establishing a licence-free Python and browser/GPU processing foundation that informed subsequent web platforms.
Reutech Mining / Reutech Radar Systems — Stellenbosch, South Africa circa 2010–2011
Role: Software Developer | System Architect
Stack: C# · ASP.NET · Silverlight · WinForms · Entity Framework · SQL Server · Java · Telerik UI · LINQ · Regular Expressions · SMTP · FTP · IIS

A family of distributed monitoring and diagnostic applications supporting a global fleet of radar systems. The software collected operational telemetry, interpreted large volumes of log data and provided centralised health monitoring and detailed investigation tools for technical support.

Key Contributions
  • Architected a distributed telemetry platform that collected log deltas from globally deployed radar systems and parsed large volumes of unstructured, hard-to-interpret log data into structured operational events.
  • Built centralised monitoring dashboards with real-time status aggregation across radars and subsystems, providing drill-down views, filtering and visual health indicators for technical support.
  • Developed a standalone forensic analysis application capable of parsing months of radar logs directly into memory, with custom zoomable event timelines, interactive charts and detailed contextual drill-down.
  • Designed recovery and continuation mechanisms across application and virtual-machine failures, and partitioned historical records into archive tables to preserve the responsiveness of live operational queries.
  • Delivered email alerting and automated log retrieval (FTP/SMTP) to close the loop between detection and operator response.
Outcome: Gave technical support teams worldwide real-time visibility into fleet health and the forensic tools to diagnose field issues remotely, reducing reliance on on-site investigation.
Reutech Mining / Reutech Radar Systems — Stellenbosch, South Africa circa 2007–2012
Role: Software Developer | HMI Development
Stack: Delphi · CrossKylix · OpenGL · TeeChart · SQL · TCP/IP · RS-232 · XML · Autodesk DXF

Production HMI for open-pit slope-stability monitoring radar, covering hardware configuration and control, area-of-interest drawing, live and historical trend analysis, 3D scan visualisation, 3D terrain models, georeferencing, alarms and reporting.

Key Contributions
  • Built and maintained OpenGL 3D graphics pipelines for real-time radar-scan overlays, digital terrain model (DTM) rendering and georeferenced spatial views.
  • Engineered multi-sensor fusion features, combining prism and geotechnical sensor data with radar measurements to compute and visualise displacement relative to local terrain.
  • Developed interactive time-series trends and spatial analysis tools, enabling geotechnical engineers to draw monitoring regions, track movement velocity and analyse historical slope behaviour.
  • Implemented mission-critical alarm workflows, integrating threshold-based software alerts with serial-triggered physical sirens and external warning hardware.
  • Maintained and evolved a long-lived, multi-OS production codebase (Windows/Linux via CrossKylix), ensuring continuous 24/7 operational stability in harsh mining environments.
Outcome: Modernised the monitoring workflow by replacing legacy 2D views with an interactive 3D terrain and radar environment, delivering reliable 24/7 slope-monitoring and critical early-warning capabilities to open-pit mining operations worldwide.
Azilla Systems (now ODEX) — Cape Town, South Africa circa 2004–2005
Role: Web Application Developer
Stack: HTML/CSS · JavaScript · AJAX · SVG · GIS · ASP · SQL Server · SQL Server Reporting Services · IIS · XML/SOAP

Geospatial and floorplan visualisation module built as part of a multi-tenant SaaS platform, enabling outdoor advertising operators to explore, package and manage national billboard and venue inventory.

Key Contributions
  • Independently developed an interactive GIS module integrating Esri ArcGIS with geospatial search, points of interest, demographic data and advertising inventory.
  • Implemented spatial plotting and clustering to make large, geographically distributed billboard inventories intuitive to explore and navigate.
  • Developed vector and raster floorplan visualisations (zooming, panning, navigation) linked directly to underlying asset and client data.
  • Built dynamic, desktop-like browser interfaces using early AJAX techniques to provide seamless controls and data views without full-page reloads.
Outcome: Turned geographic and floorplan data into an interactive visual component of a production campaign-management platform.