Skip to content
View SzematPro's full-sized avatar

Block or report SzematPro

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SzematPro/README.md

Waldemar Szemat

AI Solutions Architect · Fractional Head of AI · Santiago, Chile · Remote


Most GenAI pilots never make it to production. Getting them there and keeping them running at scale is what I do.

As CTO of a startup in a regulated industry, I built a product from zero with generative AI at its core and scaled it to thousands of users. The hard part was never the model. It was everything around it: the integrations, day-to-day operations, and an architecture that stayed stable at scale.

Today I run scoped POC-to-Production Sprints for teams whose GenAI project is stuck, with a focus on regulated industries (finance, insurance, public sector, healthcare): I come in, find why it isn't shipping (it's rarely the model), and get it there. Plus advisory on designing scalable AI systems.

How I work

  • Spec-driven development: clear specs first, then build, so AI handles the mechanical work and humans make the high-order calls
  • Multi-agent systems & GenAI in production: agents, RAG, evals, guardrails. Built for production, not demos
  • Multi-stack, vendor-agnostic: Azure AI, Anthropic, LangGraph, Semantic Kernel, OpenAI, plus AWS and GCP. I pick the stack that fits the problem

I work both sides: the commercial (tech consulting, team management, client-facing) and the technical (engineering, architecture, hands-on in the code). Full-stack engineer, technical lead, then CTO. EN/ES.


Flagship: an evaluated AI agent for regulated industries

A conversational, cite-or-refuse AI agent built for a regulated environment (healthcare as the chosen instance: the highest bar), paired with a CI-gated evaluation harness: eight scorers, five deterministic CI gates, three judge-backed. Trilingual (EN/ES/PT-BR), built and evaluated on 100% synthetic data; a capability and readiness reference, not a medical device.

Elsewhere

szemat.pro · LinkedIn · Microsoft Learn · [email protected]

@SzematPro's activity is private