Jay Vaghela
AI Engineer · Vadodara, India
I build production AI systems for environments where "the model hallucinated" is not an acceptable post-mortem — multi-tenant payments, regulated enterprise data, real-time voice agents.
The part I find most interesting isn't the model itself; it's everything around the model that makes it behave: retrieval that returns the right thing, agents that know when to stop calling tools, voice that holds a real conversation over a real phone line, and safety boundaries that survive the LLM trying to be helpful at the wrong moment.
Around 5 years writing software, last 3 mostly on GenAI — long enough to remember when "agentic" was just a typo. Currently lead the AI team at AtliQ Technologies.
Selected work
Payments Central — AI Assistant
AI assistant embedded in an institutional cross-border payments platform. Handles documentation and live operational questions across multi-tenant client portals, with strict tenant isolation across every AI surface.
Voice AI Telephony Agent
Outbound calling agent sustaining 300+ sales-qualification conversations per day. Real-time turn-taking on noisy lines, seamless human hand-off, full conversation-safety layer.
ragdb — Natural-Language-to-SQL Package
Internal package giving non-technical users plain-English access to any SQL database. Engine- and provider-agnostic. Powers the live-data layer across multiple AtliQ AI engagements, including Payments Central.
KnoGen — Enterprise Agentic AI Platform
Agentic AI platform for querying organizational knowledge bases through natural language. Role-based access, audit logging, and multi-provider model routing for regulated deployments.
Stack
Python · TypeScript · FastAPI · Node.js · PostgreSQL · React Native · LangChain · LangGraph · LlamaIndex · LiveKit · OpenAI · Anthropic · Mistral · Gemini · Docker · AWS