Kurt Overmier
AI Product Engineer | AI Systems & Automation Leader
San Antonio, TX · (512) 450-4801 · kurt@kurtovermier.com · GitHub · NPMProfessional Summary
AI product engineer and systems leader building practical, production-oriented AI workflows with approximately 2.5 years of hands-on engineering experience and 20+ years in operations and people management. Design and ship multi-agent platforms, model-routing services, RAG applications, governance tooling, and human-in-the-loop workflows. Bring an operator’s perspective to reliability, adoption, measurement, and the points where automation must hand work back to people.
Core Expertise
- AI & Agentic Systems
- Multi-LLM orchestration and routing, RAG, embeddings, vector search, tool calling, structured outputs, agentic workflows, human-in-the-loop escalation, evaluation-oriented prompt iteration, AI governance
- Software & Cloud
- TypeScript, JavaScript, React, Node.js, Hono, FastAPI, Python, SQL, REST APIs, webhooks, Cloudflare Workers, Durable Objects, D1, KV, R2, Queues, Vectorize
- Product & Operations
- Workflow design, requirements, rollout and adoption, SLA/KPI ownership, quality systems, dashboards, forecasting, escalation analysis, cross-functional delivery
- Data & Platforms
- Qdrant, Power BI, Looker, Looker Studio, Salesforce, Zendesk, ServiceNow, Jira, Qualtrics, Excel, Google Sheets
Professional Experience
SmartBrandStrategies / StackBilt — Founder & AI Platform Architect
- Build and ship production-oriented multi-agent AI systems: define agent roles and scopes, orchestrate handoffs, route across LLM providers, integrate tools, and design human-in-the-loop escalation paths.
- Built an AI governance and decision-support system for decision logging, standards validation, audit trails, and operational-drift reduction across multi-agent workflows.
- Integrated OpenAI, Anthropic Claude, Groq, and Gemini-style providers with model routing, structured prompts, tool workflows, and retrieval-augmented generation patterns.
- Built semantic retrieval workflows with Cloudflare Vectorize and Qdrant; architected cloud-native systems using Workers, Durable Objects, D1, KV, R2, Queues, Hono, TypeScript, React, Node.js, FastAPI, Python, and SQL.
- Ship public NPM packages and developer tooling, including the Charter CLI for modular AI-agent context management and governance.
OneSupport — Operations Manager / AI Systems & Automation Lead
- Lead remote and distributed customer-support teams of 30–45 agents, owning SLA/KPI performance, staffing, coaching, quality, performance management, and escalation trends.
- Designed and deployed a QA automation pipeline that reduced quality-review time approximately 70%, carrying the work from operational need through requirements, implementation, rollout, adoption, and measurement.
- Coordinated with IT on an RLHF-powered knowledge assistant built from 2,100+ curated Q&A pairs; owned requirements, data curation, rollout, and frontline adoption.
- Build reporting workflows in Python, SQL, Power BI, Looker, and Looker Studio to turn CSAT, NPS, AHT, utilization, quality, and escalation signals into operating decisions.
TaskUS — Operations Manager
- Led 450 FTEs and 10 direct-reporting managers during the Disney+ launch; data-driven coaching, QA dashboards, and operational reporting improved CSAT 23% and earned preferred-partner status.
OneSupport — Social Media & Operations Manager, Google Fiber Support
- Managed a 30+ agent operation across voice, email, and social channels for Google Fiber support.
- Partnered with Google engineering stakeholders to build automation and social-listening workflows that reduced multi-channel response time from 26+ hours to under 10 minutes.
- Built workforce-forecasting models with approximately 92% accuracy, enabling proactive staffing and capacity decisions.
Education
Texas State University — B.B.A. in Marketing | San Marcos, TX | 1997–2003
Additional Information
Languages: English (native); Spanish (conversational)
Public proof of work: github.com/stackbilt-dev | npmjs.com/~stackbilt-kovermier