A solo studio of one — building chatbots, voicebots & retrieval systems for banks, hotels and contact centres. Eight Rag services running right now on one local machine, behind one nginx.
No managed cloud. Every service in this portfolio runs on one host I administer myself, behind a single nginx gateway. Cheaper to run, quicker to change.
No vendors. Local LLM inference for the parts that matter, with Claude & Gemini routed through local API adapters I control end-to-end.
One team — one human. From Genesys flow to voice pipeline to retrieval index, I write & ship every layer. Fewer handoffs, faster iteration.
Four years building conversational systems end-to-end. Every line — flow logic, LLM glue, voice pipeline, retrieval index — written by one person, shipped to production.
Four short steps from a conversation about your problem to a service running on your domain. Async, transparent, no hand-offs.
30-minute call. Your goal, your constraints, the existing CCaaS stack & where conversational AI actually moves the needle.
~ 1 dayScope, milestones & a shared repo. I spin up the local stack, wire the model gateway, and we start in days, not months.
~ 1 weekWeekly demos against real conversations. Iterate the prompts, flow, retrieval & voice pipeline until the metrics actually move.
~ 4 weeksGo-live on your domain or mine. Ongoing tuning, observability, model upgrades — the service grows with the conversation volume.
ongoingA small, opinionated set of tools that fits on one machine — chosen because I can read every line of every layer when something breaks.
I work in Python·Genesys Cloud— and lately —LLM fine‑tuning·QLoRA·agentic systems·retrieval.
Drop a line. I reply within 24 hours and the first call is free. If we don't fit, I'll point you toward someone who does.