Christi Akinwumi
I design, ship, and measure conversational AI.
Dallas-Fort Worth, TX · christi.io · [email protected] · linkedin.com/in/christiakinwumi · github.com/mrskinzo
Highlights
- Shipped conversational AI to 20M+ users across 8+ international markets.
- Led an enterprise migration from rule based NLU to an LLM multi-agent system: 25% faster responses, $2M annual support savings, 60% fewer hallucinations.
- Built and operate a live tool-using booking agent for a real business, on autopilot since launch (boothlyevents.com).
Experience
Principal & Founder · Intelligent CX Consulting, LLC
Dallas-Fort Worth, TX (remote, US clients)
- Run conversational AI consulting engagements: diagnostic sprints, eval harness builds, intent taxonomies, and orchestration architecture for support organizations.
- Designed and shipped Boothly, a tool-using Claude agent for a photo booth business. It triages leads, builds quotes from a single pricing source, places holds on the real Outlook calendar, and confirms by SMS. Lead to quote, hold, and text in under 60 seconds, 24/7.
- Rebuilt planning and review rituals for a cross market AI product team and onboarded 200+ non technical employees to agentic AI tooling.
Conversation Designer · GoDaddy
- Led the migration from rule based NLU to an LLM multi-agent support system across 8+ markets and 20M+ users: 25% faster response time, $2M annual savings, 40% CSAT lift.
- Rebuilt the routing taxonomy from 50K+ real conversations: 150+ mutually exclusive intents across 12 product domains, misrouting down 30%, handle time down 20%.
- Designed and ran a six week RAG evaluation across three configurations. The winning hybrid retrieval architecture shipped to production: 45% accuracy lift, 60% fewer hallucinations.
- Authored the vendor decision framework (OpenAI, Anthropic, AWS Bedrock) that secured a $500K+ budget and leadership buy in.
Workflow Author · Peloton
- Rebuilt self service for a connected fitness product across 8 international markets: information architecture, decision trees, localization, and a product gap feedback loop that fed the roadmap.
- Resolution time down 40%, self service adoption up 35%, support volume down 20%.
Selected projects
- Boothly Booking Concierge. Live tool-use agent for a real business: Claude (Anthropic SDK), Microsoft Graph, Twilio. Case study at christi.io/work/boothly-booking-concierge.
- RAG Pipeline Comparison. Public companion repo to the production RAG evaluation: three configurations, one eval harness. github.com/mrskinzo/rag-pipeline-comparison.
- christi.io. This site: Astro, a custom build pipeline, and a Claude powered chatbot grounded in my published writing.
Education
- M.S., Curriculum & Instruction
Texas A&M International University - Post Graduate Certification, Technical Communication
University of North Texas - B.G.S., Business
Texas Woman's University
Certifications
Skills
- Conversation design
- Multi-agent orchestration
- Prompt architecture
- RAG evaluation
- Eval harness design
- Intent taxonomy
- LangFuse
- Dialogflow CX
- Azure AI Foundry
- AWS Bedrock
- Claude (Anthropic)
- JavaScript
- Python