What I can build for you
One-time engineering projects, dataset purchases and recurring agreements. Pick the one that fits, or describe your project and I'll tell you which fits.
Custom AI Voice Persona Engineering
$200–$800 standard · $1,000+ larger projects
An AI personality with its own voice, character, conversational style and behaviour rules, connected to a live speech pipeline so people can actually talk to it. This is the same work behind Rebecca, my flagship persona.
Get a quoteFor AI VTubers, streamers, content creators, gaming communities, startups and businesses that want an interactive AI personality.
- Personality and character design
- Voice configuration and text-to-speech integration
- LLM configuration and conversational behaviour
- System instructions and persona development
- Speech-to-text → LLM → text-to-speech pipeline integration
- Real-time voice interaction setup
- Testing, troubleshooting and refinement
- Integration with supported applications and environments
AI Training Data & Model Training
Datasets from $0.45/row ($1,000 min) · data prep from ~$400 · project quotes
Three levels, depending on where you're starting from: buy a prepared dataset, have me turn your own raw data into a trained and evaluated model, or keep your model improving every month.
Compare the tiersFor AI startups, independent developers and teams that need prepared training data, custom cleaning, fine-tuning or evaluation.
Ready-to-use datasets
Cleaned, structured, validated and documented, with an optional training upgrade.
Custom data prep + training
Your raw data → training-ready dataset → fine-tuned, evaluated model.
Managed monthly training
Scheduled preparation, training and regression testing on a defined scope.
Managed Monthly Model Training
Starter $1,000–$2,000 · Growth $2,500–$5,000 · Premium $5,000–$10,000+ /mo
Ongoing model improvement on a defined monthly agreement, at a better effective rate than buying the same work project by project. Each agreement sets out data volume, training frequency, evaluation, deliverables, turnaround and support limits.
See the plans- Agreed volumes of new or updated data processed
- Automated cleaning and structuring with human quality review
- Dataset validation and preparation
- Scheduled training or fine-tuning jobs
- Evaluation against established benchmarks
- Comparison of each new version against the previous one
- Updated model versions delivered
- Performance reports covering improvements, regressions and limitations
Real-Time Voice AI Engineering & Optimization
Custom quote · larger projects from ~$1,000
For teams building their own conversational voice product. I work on the parts that make real-time voice feel natural or broken: how fast it answers, whether it can be interrupted, whether it knows when it's its turn, and how human it sounds.
Get a quote- STT, LLM and TTS pipeline configuration
- Real-time voice AI integration
- Latency and responsiveness improvements
- Speech interruption (barge-in) and turn-taking refinement
- Conversational behaviour testing
- Voice quality and naturalness testing
- Interaction evaluation and troubleshooting
- Integration with supported applications and environments
Indicative ranges
| Service | Indicative price | Priced by |
|---|---|---|
| Custom AI voice persona | $200–$800 | Persona depth, voice work, integration target |
| Larger custom engineering | $1,000+ | Scope |
| Ready-to-use dataset | $0.45/row · $1,000 min | Row count ($0.30/row above 10,000), non-exclusive licence; updates $300/mo |
| Dataset + model training | Package quote | Training and evaluation requirements |
| Data preparation only | ~$400–$1,000 | Volume and messiness of the data |
| Custom data prep + training | Project quote | Data, model, training runs, evaluation |
| Managed monthly training | $1,000–$10,000+/mo | Volume, run frequency, evaluation, reporting |
| Real-time voice AI engineering | Custom quote | Complexity |
These are indicative ranges, not fixed prices. Final pricing depends on data volume, complexity, model compatibility, training frequency, evaluation needs, delivery requirements and how much human quality review is needed. AI API usage, hosting, GPU compute, storage and other infrastructure costs are billed to the client separately unless the agreement explicitly includes them.