Services

1. Use-case audit

60–90 min review of your workflow, data, and constraints. Deliverable: written recommendation — build locally, use API, or do nothing.

2. Hardware + cost plan

GPU / CPU / RAM / storage sizing for your models and load. Includes purchase list, power estimates, and 12–24 month TCO vs. hosted APIs.

• Single workstation to small cluster
• Mac, NVIDIA, or CPU-only options
• Offline / air-gapped considerations

3. Local deployment

Install and configure open models, embeddings, and retrieval over your docs. Private chat / API endpoints your team can use.

• Ollama / llama.cpp / vLLM / text-generation-webui as appropriate
• RAG over PDFs, wikis, tickets
• Basic eval + guardrails

4. Handoff + support

Runbooks, update procedure, and backup checklist. Optional follow-up block for tuning.