Here is the blueprint of how we engage. Our approach is the result of distilling experience from a wide variety of completed projects.
Our workflow combines extensive data exploration with well-designed experiments and a constant feedback loop with your business experts. A typical full engagement runs 6 to 8 weeks. Larger projects can be scoped into multiple steps.
Timeline
Phase 1 — Evaluation
We talk about your goals, preferably in person. R&D work starts once a usable compute environment is ready and your data is accessible within it.
Phase 2 — Discovery
We do a deep dive into the data, metadata, and your business requirements. The goal is to check feasibility and develop a sound Statement of Work (SOW) with a realistic timeline, milestones, success metrics, and stop/go decision points. We typically work on an hourly rate during this phase.
Phase 3 — Execution
We work as outlined in the SOW with weekly touchpoints. We include your subject matter experts as needed, but won’t fill your calendar with meetings — status updates and discussions happen in a shared project workspace that keeps everything in one place and tracks progress for everyone. We often switch to milestone-based pricing for this phase.
Phase 4 — Knowledge Transfer
We collect, document, and hand over all artifacts: a well-documented code repository, a summary report, and a final presentation for stakeholders. We support your team in deploying models and integrating the tools we developed into your workflow, and discuss next steps and future opportunities.
Technical Infrastructure
On-Premise
- Ubuntu LTS (latest) with developer tools
- Python 3.11+
- 64 GB+ RAM
- NVIDIA GPU with 24 GB+ VRAM
Cloud
- AWS
- Google Cloud
- Microsoft Azure
- DigitalOcean