Chris Dabatos

About Chris Dabatos

I believe the best way to reach developers is to build things they’d actually build themselves.Over the last few months I’ve shipped 8 AI apps, each technically more intricate than the last.My latest OSS app is RecallMEM, an open source deterministic memory framework for LLMs. It’s been published to NPM with 900+ weekly downloads and a hosted cloud version launching soon.This app was heavily influenced by my earlier project, Speak2Me.io, which also solves the challenge of giving AI persistent and reliable memory across conversations.It’s a Voice AI Agent with a three-tier memory architecture that never loses a fact about you. It uses real-time voice with emotion detection, deterministic memory on top of vector search, and a study mode that does RAG over any knowledge base and answers questions in real-time via voice.I wrote about the architecture here: currently exploring how these same patterns can apply to enterprise AI at scale.

Experience

1. Developer Relations - Ai & Data Infrastructure

   TiDB, powered by PingCAP

   Jul 2024 — Present · Sunnyvale, CA, US

   My goal is simple: help devs get comfortable using AI in their workflow, not just talk about it- Design and build tool-calling agent architectures using Claude, MCP, and TiDB\'s vector search APIs. Built BaseCamp, a voice-first health OS where an LLM agent decides when to call external tools (search_food, scale_nutrition) to interact with real data systems. This is the exact pattern MCP formalizes for agent-to-tool communication- Led the TiDB AgentX Hackathon 2025 to record-breaking scale with 2,700+ developers across 95 countries. Personally reviewed the code of 219 submissions, identified adoption friction points in docs and SDKs, and drove a 45% increase in TiDB Serverless cluster adoption- Built Speak It, a voice-to-text Chrome extension that learns speaking style through statistical fingerprinting without storing raw content, applying data minimization and privacy-by-design principles to AI-powered tooling- Engineer responsible AI practices into every demo: hallucination detection that tracks whether responses come from verified data or LLM generation, per-request cost transparency, rate limiting and abuse protection for public-facing AI apps, and safe rollback workflows- Produce technical content across YouTube, documentation, webinars, and conference stages (Big Data LDN, Percona Live, OSACON, All Things AI), all derived from building real applications, not theoretical walkthroughs.

Skills

• Json
• Jquery
• Html5
• Javascript
• Sales
• Social Media Marketing
• Customer Service
• Ajax
• Management
• React.js
• Account Management
• Event Management
• Cascading Style Sheets (Css)
• Bootstrap3
• Customer Satisfaction