Services
Consultancy & R&D in Audio Software and On-Device AI
MAP Audio is the product studio behind PAM and MFX. Services put the same team on your product — DSP, audio plugins, on-device AI, and the release infrastructure around them — whether that means a first spec, a feature, or a legacy codebase nobody wants to touch. If you think we can help, we want to hear from you.
C++20 · JUCE · MAE · React · llama.cpp · MLX · ONNX Runtime · Core ML
What We Do
On-Device AI
Models shipped inside your product
We take a model from checkpoint to a feature running on your users’ machines: fine-tuned, quantized, offline, and wired into a real product’s threading and lifecycle rules.
- Small-model fine-tuning: datasets, training, eval harnesses
- llama.cpp, MLX, ONNX Runtime, and Core ML runtimes
- Generative image/video workflows: diffusion pipelines, ComfyUI, prompt systems
- Inference ops: GPU serving, model deployment, cost and performance tuning
- Real-time-safe integration: inference output drained into the audio thread over lock-free queues
- Model download, verification, and version pinning
Audio Plugin Engineering
From idea to market-ready plugin
Complete plugin development in modern C++: VST3, AU, CLAP, and standalone from one codebase, for macOS and Windows.
- JUCE engines with React UIs in a native WebView
- Samplers, granular engines, loopers, sequencers
- Audio-reactive visual engines and A/V capture/export pipelines
- Hosting third-party VST3/AU/CLAP plugins inside your product
- Multi-product builds: several SKUs from one source tree
DSP Design & Implementation
Original algorithms under production constraints
Signal processing written for the audio thread: allocation-free, branch-light, regression-locked with golden render tests.
- Saturation and wavefolding with ADAA antialiasing, no oversampling
- Time-stretching: phase vocoder, transient-slice, and granular engines
- Drum synthesis, resonator banks, tape and character effects
- Golden-hash test suites that pin every algorithm change
Codebase Modernization
Legacy audio code, brought current
Codebases other teams are afraid to touch: we migrate them to modern C++, make the audio path real-time safe, and leave tests that keep it that way.
- Migration to C++20 without a rewrite
- Real-time-safety audits: locks, allocations, and syscalls off the audio thread
- Characterization and golden-render tests around untested DSP
- Performance tuning backed by profiler traces
AI-Agent Integration
Your product, drivable by AI agents
Agents can operate your product the way a user does: load content, change parameters, run whole sessions, live. We build that bridge into the running software and design the tool surface so agents drive it reliably.
- Live bridges into running desktop software
- Tool-surface design: schemas, error messages, and docs written for agents
- MCP servers over your product’s API
- In-product assistants backed by your own state and documentation
Agentic Development
Agent-first codebases and teams
We make your codebase and processes agent-first, and onboard engineering and product teams to advanced agentic development.
- Agent-first codebase setup: project rules, code maps, and agent skills
- Onboarding engineering and product teams to agentic workflows
- Scheduled agents for maintenance and regression watch
- Agent-run QA against the live product over an MCP bridge
Licensing, Store & Release Infrastructure
Everything around the binary
The machinery that makes a product sellable: licensing, delivery, installers, signing, and the pipelines that validate a release before users see it.
- License activation and trial flows
- Storefront and content delivery on Cloudflare Workers and R2
- Installers, code-signing, and notarization for macOS and Windows
- Release validation pipelines that catch a bad build before it ships
Advisory & Strategy
Senior guidance without the full-time hire
A standing advisor for teams building audio and AI products. We help you decide what to build, what to skip, and keep the plan honest as it meets reality.
- Technology roadmaps: platform, framework, and build-vs-buy decisions
- Research roadmaps: which ideas earn a prototype
- Quarterly reviews of architecture, codebase, and process
- Team workshops on audio engineering and agentic development
How We Work
You talk to the engineer writing the code. Three ways to do it.
Fixed-scope build
A feature or a whole product, end-to-end: spec, milestones, code, tests, handover.
Embedded engineering
A senior engineer inside your team on a monthly retainer. Your repo, your process, work merged weekly.
Audit & rescue
A review of a codebase, its performance, or its real-time safety. Written report, prioritized fix plan. Also the first step of any modernization job.
The Team
A small team with one foot in research. These are the people you work with.
Jack Armitage
Research Engineer & Musician
Founder of an art-science studiolab in North Iceland. Over 30 publications as a postdoc at the Intelligent Instruments Lab in Reykjavík, and PhD at the Augmented Instruments Lab in London. Former research engineer at ROLI and FXpansion. Production credits include Charli XCX and Jónsi. Live coding as Lil Data on PC Music for >10 years.
jackarmitage.com →Can Ince
Founder · Principal Engineer
Former FL Studio developer at Image-Line and lead developer at Lunacy Audio; interface research published at NIME and ICMC, MA from the University of Huddersfield. Designs and develops instruments end to end: PAM’s sampler engine, DSP, UI, on-device AI, and the release infrastructure behind it.
ince.io →Mert Toka
AI Research Engineer
Co-founder of Metaesthetica, a research-design collective for immersive media experiences. Researcher in generative systems and human-computer interaction, published at ACM CHI, DIS, and SIGGRAPH Art Gallery; PhD from UCSB as a Fulbright scholar. Works across audiovisual systems and computational fabrication.
merttoka.com →Talk to Us
Tell us what you’re building and where it’s stuck. A short email is fine.