ApotheCode Pty Ltd
ApotheCode is an Australia-based software engineering and R&D consultancy. We work with businesses, research groups, and development teams on systems where performance, correctness, or domain complexity make off-the-shelf approaches inadequate.
Australia-based. All work is delivered directly by the founding director.
Technical focus areas
Services
Engagements are scoped to the problem, not sold as fixed packages. These categories reflect the main classes of work ApotheCode undertakes.
Architecture, implementation, and debugging for systems where concurrency, distributed state, or performance constraints make correctness non-trivial.
Problem types
Simulation pipelines, GPU-accelerated computation, and numerical modelling for research groups and engineering teams with scientific or computational research requirements.
Problem types
Unreal Engine and Unity systems work focused on technical architecture: C++ plugins, procedural generation, and in-engine simulation — not game design.
Problem types
Bespoke software for businesses where off-the-shelf products do not fit: internal tools, access-controlled backends, and domain-specific applications.
Problem types
Not sure which category fits? Most real problems cross multiple domains. Describe the problem briefly via the contact form and we will advise whether it appears to be in scope and which engagement model is most suitable.
Capability
All work is delivered directly by the founding director. There is no delegation to junior staff or outsourced delivery.
Principal-led
Mark Devereux
Founding Director & Principal Software Engineer
Formal training in neuroscience, pharmacy, clinical pharmacy, and information technology enables direct engagement with technically demanding problems in scientific, simulation, and regulated-domain software.
Academic Background
Systems Domains
Technology Stack
Languages
Frameworks & Runtimes
Parallel & GPU Computing
Engines
Infrastructure
Data
This combination of domain training and software engineering practice is particularly useful in simulation, scientific tooling, and systems that must be specified carefully before they can be implemented correctly.
Services — Detail
Each service area has a defined scope, typical delivery types, and technical constraints that shape how engagements are structured.
Software where correctness matters beyond basic functionality. This covers systems where architecture decisions have long-term operational cost, concurrency introduces failure modes, or debugging requires systematic analysis rather than iteration by guesswork.
Problem types
Delivery types
Key constraints
Typically involves trade-off analysis between correctness, performance, and maintainability. Deployment targets (cloud provider, container runtime, CI/CD pipeline) affect design decisions significantly.
Computational systems for research groups and engineering teams whose work requires both domain accuracy and production-quality software. Covers simulation pipelines, physical and pharmacological modelling, and numerical computation at scale.
Problem types
Delivery types
Key constraints
Domain accuracy is a hard constraint. Performance requirements often dictate the compute model. Hardware availability (GPU VRAM, cluster topology) shapes what is feasible.
Unreal Engine and Unity development with a systems engineering focus. This is not general game design work — it covers the technical substrate: engine systems, plugins, procedural generation, and performance-critical runtime code.
Problem types
Delivery types
Key constraints
Unreal Engine's threading model imposes hard constraints on what can run off the game thread. Plugin compatibility across engine versions must be considered. Memory budgets are often tighter than they appear.
Custom software for businesses whose requirements cannot be met by an existing product. Covers internal tooling, access-controlled backends, and domain-specific applications. Design emphasis is on correctness and long-term maintainability, not speculative scale.
Problem types
Delivery types
Key constraints
Bespoke software requires detailed specification to scope correctly. The main scoping risk is incomplete problem definition at the outset. Engagement typically starts with a structured requirements phase.
Selected Work
Summaries are intentionally limited to non-confidential technical detail. Ongoing and private work is labelled accordingly.
Context
Internal R&D work for a large-scale procedural world generation system requiring processing of high-resolution climate and biome datasets.
Problem
Processing over 700 million climate and biome data points within a feasible timeframe using consumer-grade GPU hardware.
Technical Constraints
Consumer GPU VRAM limits imposed hard memory budgets across the processing pipeline. CUDA kernel design required coalesced access patterns to avoid memory bandwidth as the bottleneck. Host/device transfer overhead required explicit staging to remain within viable throughput.
Approach & Status
CUDA-based parallel processing pipeline with tiled memory access, batched data staging, and C++ host-side orchestration. Kernel design optimised for memory throughput rather than compute density.
In development. Initial architecture and throughput approach have been validated against target dataset scale.
Context
Component of the same procedural world generation system, implemented as an Unreal Engine C++ system.
Problem
Generating large-scale biome maps at runtime without blocking the game thread or introducing non-determinism.
Technical Constraints
Unreal Engine's threading model requires careful use of the TaskGraph and async systems. Blueprint exposure required stable API design. Deterministic output required fixed-seed generation logic.
Approach & Status
C++ multithreaded architecture using UE TaskGraph. Biome parameters exposed to Blueprint for designer control. Deterministic generation via seeded noise and reproducible traversal order.
In development as part of broader procedural generation system.
Context
Commercial application for remote monitoring workflows with configurable, multi-role access control requirements.
Problem
Required a correctly-gated backend deployable to both Vercel edge and self-hosted Kubernetes without maintaining separate codebases for each target.
Technical Constraints
RBAC logic had to be consistent across all API routes without route-level duplication. Docker and Kubernetes configuration had to remain compatible with Vercel serverless deployment constraints.
Approach & Status
MERN stack backend with structured RBAC middleware applied at the route level. Containerised with Docker, orchestrated with Kubernetes for self-hosted deployment. Vercel-compatible configuration for serverless targets.
Backend systems in active development.
Context
Simulation plugin for educational and research use within Unreal Engine, modelling physiological and pharmacological systems.
Problem
Implementing domain-accurate models of physiological and pharmacological processes within the constraints of a real-time game engine environment.
Technical Constraints
Biological model accuracy is a hard constraint. Simulation update rates needed to remain stable under engine frame budget. Plugin architecture required clean separation from game logic.
Approach & Status
Unreal Engine plugin with computational models grounded in the underlying physiological and pharmacological domain. Update scheduling decoupled from render frame rate.
Selected R&D work. Domain validation ongoing.
Context
Android application providing structured, searchable access to materia medica reference data for professional or research use.
Problem
Making a large, structured dataset navigable and searchable on Android without relying on a live backend dependency.
Technical Constraints
Data accuracy and source integrity. Offline-first operation required. Android SDK constraints on local data storage and search performance.
Approach & Status
Native Android application with structured local data layer, SQLite-backed search, and clean information hierarchy.
Selected project work.
Commercially sensitive or ongoing client work is not listed. Additional project detail may be discussed during a relevant engagement conversation, subject to confidentiality constraints.
Engagement Model
Four structured engagement types for different problem sizes and time horizons.
All engagements begin with a brief intake — either a short form submission or a direct booking. Initial contact is used to assess scope and fit rather than route enquiries through a sales process.
Initial consultations are used to assess scope and fit. If the work is outside the consultancy's capability, that will be stated clearly.
A structured diagnostic session for a defined technical problem. The output is a structured assessment of the likely root cause, the main uncertainties, and the recommended next step.
Suitable for
Structured review of existing code, architecture, or system design. Deliverable is a written assessment with prioritised findings and specific recommendations.
Suitable for
A defined engagement with a clear deliverable: new system, subsystem, integration, or migration. Structured as proposal, delivery, and handover.
Suitable for
Retained technical input across an extended project or series of workstreams. Used where consistent involvement is preferable to one-off delivery — iterative R&D, ongoing architecture decisions, or sustained specialist support.
Suitable for
About
ApotheCode Pty Ltd is a principal-led software engineering and R&D consultancy based in Australia.
ApotheCode works on systems where the main difficulty is not writing code, but structuring the problem correctly. This includes concurrent and distributed systems, GPU-accelerated computation, simulation and modelling, game engine systems, and bespoke business software.
The consultancy does not attempt to serve every software need. Engagements are accepted where the problem type is a clear match for the underlying capability.
Where a problem falls outside the consultancy's expertise, this is stated directly rather than deferred.
Founding Director
Mark Devereux
Founding Director & Principal Software Engineer
Brings formal training in neuroscience, pharmacy, clinical pharmacy, and information technology, supporting work at the intersection of scientific modelling, computational systems, and software engineering.
Technical practice spans systems-level C++ and CUDA, full-stack web and mobile development, game engine work in Unreal Engine and Unity, and deployment and infrastructure work with Docker, Kubernetes, and cloud platforms.
This multidisciplinary background is particularly useful where technical implementation depends on understanding the underlying model, especially in simulation, physiological modelling, and research software.
Positioning
Contact
Use this form to describe the problem or engagement you have in mind.
Prefer to schedule directly?
Request a consultation to discuss your project or problem.
Request a ConsultationDirect email
contact@apothecode.com.au