Hello, I'm

Enrique Mascarua

|

Building production-grade RAG, multi-agent and voice AI systems — not demos. 5+ years shipping LLM platforms across AWS, Azure and GCP.

About Me

Enrique Mascarua Ley

I started at 18 as CTO of my own AI startup building RAG and agentic systems from scratch. Five years and four companies later, I lead a team of 6 engineers at NexOne AI, architecting financial document intelligence and conversational AI platforms. My stack: LangGraph, LangSmith, RAGAS, Snowflake, AWS, Retell AI, VAPI and Twilio SIP — focused on shipping LLMs that survive production traffic, not notebook demos.

5+
Years in Production AI
6
Engineers Led
4
AI Companies
3
Cloud Platforms

What I Build

Three areas where I've shipped systems that survive real users and real traffic.

Production RAG & Multi-Agent

LangGraph orchestration, GraphRAG retrieval, evaluation pipelines with RAGAS and LangSmith for measurable quality.

Voice AI Systems

Real-time voice pipelines with Retell AI, VAPI and Twilio SIP. Custom TensorFlow STT for domain-specific accuracy.

LLM Observability & Cost

Tracing with LangSmith, Datadog dashboards and token-cost optimization to keep LLM features economically viable.

Experience

Sr AI Engineer

Apr 2025 – Present

NexOne AI

Leading end-to-end development of AI-driven financial document intelligence and conversational platforms. Directing a team of 6 engineers across LangGraph, Snowflake, AWS and Azure.

LangGraphLangSmithSnowflakeAWSAzureRAGAS

Gen AI / Voice AI Engineer

Dec 2024 – Apr 2025

Vocal AI

Built Twilio SIP voice pipelines with custom TensorFlow STT models for automated real estate client interactions across inbound and outbound flows.

Twilio SIPTensorFlowVoice AIPythonRetell

SWE / AI-ML Engineer

Jan 2024 – Nov 2024

Gin Real Estate

Developed agentic AI workflows with LangChain and Snowflake for property data analysis. Integrated Meta Cloud API for WhatsApp lead automation.

LangChainSnowflakeAWSMeta APITwilio

Software Engineer / Prompt Engineer

Oct 2022 – Jan 2024

Codey

Built Azure Functions for SWIFT payment processing. Designed prompt engineering pipelines integrated with Microsoft Dynamics CRM.

Azure FunctionsLangChainSWIFTDynamics 365

AI/ML Engineer & CTO

Jan 2020 – Jun 2022

Mesurto

Co-founded and led technical development of an AI startup. Built RAG pipelines and agentic AI systems from the ground up on AWS.

LangChainAWSAgentic AIRAGStartup

Skills & Technologies

AI & LLM Engineering

LangChain90%
LangGraph90%
LangSmith85%
RAG / Agents95%
RAGAS80%
GraphRAG70%
Prompt Engineering90%

Voice & Conversational AI

Retell AI85%
VAPI80%
Twilio SIP90%
Custom STT (TensorFlow)75%
Meta Cloud API80%

Cloud & DevOps

AWS90%
Azure85%
GCP60%
Docker80%
Datadog75%
Vercel85%

Backend & Data

Python95%
TypeScript80%
Snowflake85%
PostgreSQL80%
MongoDB70%
Django70%

Rapid Prototyping & No-code

Lovable85%
Base4480%
n8n70%
v080%

CRM / GTM Integrations

HubSpot80%
Salesforce70%
GoHighLevel80%
InvestorFuse75%

Featured Projects

Real systems running in production — not portfolio demos.

Private

AI Financial Document Intelligence

End-to-end agentic platform at NexOne for financial document analysis and automated reporting. Orchestrated LangGraph workflows over Snowflake with RAGAS-evaluated retrieval and LangSmith tracing.

92% extraction precision~500 docs/day processedp95 latency < 2s
LangGraphLangSmithSnowflakeRAGASAWSProduction
Confidential — details on request

Voice AI Pipeline (Real Estate)

Real-time voice interaction system using Twilio SIP trunking with custom TensorFlow STT and LLM-driven dialog. Handles inbound and outbound calls with natural language understanding for real estate workflows.

~3 min avg call durationSTT WER ~8%~200 calls/day
Twilio SIPTensorFlowPythonVoice AIRetell

WhatsApp Agentic System

Conversational agent over Meta Cloud API with persistent memory and tool-use. LangGraph orchestration handles multi-turn flows, lead qualification and CRM hand-off.

~1k messages/day~15% lead conversion<2s avg response time
LangGraphMeta APIPythonAgentic AI

EAO Agent (Agentic RAG)

Production RAG pipeline with autonomous agent orchestration for document intelligence. Multi-step reasoning over vector retrieval with explicit evaluation harness.

~88% retrieval precision~4 steps per query~30% tokens saved
LangChainAWSRAGAgentic AI
Available now

Open to contract & freelance — production AI systems only

Currently taking on senior RAG, voice AI and multi-agent projects. Not demos.

RAG / Document IntelligenceVoice AI PipelinesMulti-Agent Orchestration

Let's Connect

Got a challenge that needs production AI? I'm open to senior contract roles, advisory, and interesting collaborations.

This will open your default email client.

Location

Merida, Yucatan, Mexico

Available for opportunities

Connect with me