AI engineering: agents, voice, email and RAG that work in production
Agents, voice, email and RAG that work in production.
Most AI projects stall between the demo and daily use. I build the part in between: agents that call real tools, retrieval over a company's own data, and the evaluation that tells you whether the system is actually getting answers right.
I design for cost from day one. Token usage and the financial side of running AI are part of the architecture, not something discovered on the first invoice.
Who this is for
- Businesses that want AI quietly doing repetitive work every day
- Teams with a prototype that needs to become dependable
- Products that need voice, email or chat handled by an agent that can act, not just answer
- Engineering leaders who want evaluation and guardrails built in
What it involves
- Tool-calling agents and MCP setups connected to the systems your team already uses
- RAG pipelines over your own data, with an LLM judge and evals watching quality
- Voice AI, email automation and intent understanding that route each request correctly
- Token optimization and cost design so the system stays affordable as it grows
Questions
What kinds of AI systems do you build?
Tool-calling agents, MCP setups, RAG pipelines, Voice AI, email automation, intent understanding, LLM judges and evals.
How do you know the AI is working?
Through evaluation: an LLM judge and evals run against real tasks so quality is measured rather than assumed.
Can you control AI costs?
Yes. Token optimization and the financial side of running AI are designed in from the start.