Hi, I'm

David Herrera.

AI Solutions Architect · Software Engineer

I design and ship production AI systems —RAG, MCP and MLOps— over the company's real data.

Mexico City·GMT-6

About me

Software, data & AI applied to the real world.

Software engineer building at the intersection of backend, data, and applied AI. Four years turning business problems into systems that ship, scale, and stay maintainable instead of rotting into tech debt.

Comfortable across the stack —from data model to frontend— but my current focus is RAG, MCP, and MLOps architectures over corporate data. I care that the AI systems I design are operable, observable, and debuggable in real production.

I give talks on AI and data science at universities —Tec de Monterrey, UPVM Tultitlán and TESCo— to audiences of ~200 people, closing the loop between academic theory and what actually happens once a system meets production.

Production, not demos

The interesting part isn't the tools — it's what happens once the system meets production and you have to keep it alive.

Systems you can operate

The AI I design is observable, debuggable and operable in real production — not just on the happy path of a demo.

Share what you learn

I speak at universities because closing the loop between theory and what happens in production makes all of us better.

Stack

What I build with

Backend

C#.NETPythonPostgreSQLSQL ServerJWTOAuthAPIs

Frontend

ReactTailwindTypeScriptDevExpressAstro

Cloud & Infra

AzureAWSDockerCI/CDPipelines

AI & Data

LLMsRAGMCPMLOpsPrompts

Experience

Professional journey

The interesting part of making software isn't the tools — it's what happens once the system meets production and you have to keep it alive.

BAACC Systems

2023 — Present · Mexico State · On-site

AI Solutions Architect

Full-time

Designed and shipped Connie from scratch (80% of the system): a production RAG platform serving 3 clients over their heterogeneous data —databases, PDFs, spreadsheets— with enterprise auth.
End-to-end RAG pipelines on Azure: ingestion, chunking, retrieval, and serving with prompt observability.
MCP servers connecting internal data sources to LLM clients — models consume tools without custom glue code.
Brought MLOps practices to the team: retrieval evaluation, prompt and model versioning.

LLMsRAGMCPMLOpsAzure

Fullstack Developer

Full-time

Desktop apps in .NET and Python for the shop floor — 24/7 industrial operation.
Internal web apps with .NET and DevExpress for operations workflows.
Database design and administration on SQL Server and PostgreSQL.
Android app with RFID reader for field inventory and traceability.

.NETPythonSQLAndroid

Junior Fullstack Developer

Internship

Web modules and internal APIs.
Database maintenance and technical documentation of the stack.

MVCSQLAPIs

DEC Marketing

Mar 2022 — Sep 2023 · Mexico City · Remote

Junior Web Developer

Internship

Website development and support with HTML, CSS, JavaScript and CMS (WordPress, Joomla, OpenCart).
Hosting, security certificates and domain administration.
Corporate email account management.

HTMLCSSJavaScriptWordPress

Projects

Selected work

Some of the systems I've designed and shipped — from conversational AI over enterprise data to industrial software on the shop floor.

Featured

Work / AI · 2025 · In production

Connie

From scattered knowledge to cited answers: any team queries the company's internal databases, PDFs and spreadsheets in plain language — and gets the answer in seconds.

Before: Info lost across teams or buried in red tape
My role: Architect · 80% of the system, from scratch
Scale: 3 clients in production

RAGMCPAzurePythonSQL

Work / Mobile · 2024

HandHeld RFID

An inventory that used to take weeks —or simply never happened— turned into real-time control from the handheld.

Before: Manual capture · weeks of delay
Now: Every rack RFID-tagged, every move instant
Impact: Far more accurate warehouse control

JavaAndroidSQL

Work / Desktop · 2024

Automotive Traceability

Every automotive part traced end to end across the production line.

Clients: Suppliers to Honda, GM and Stellantis
Scale: Thousands of parts per day
Scope: End-to-end traceability

C#DevExpressSQL

Personal / Web · 2026

Portfolio

This site. Editorial, bilingual and open source — hand-built design.

Type: Personal project · open source
Focus: ES/EN i18n, theming and performance
License: MIT

AstroCSSTS

Education

Learn & teach

Academic background, certifications, and speaking engagements — because knowledge grows when shared.

Academic

Information & Communication Technologies Engineering

Higher Studies Institute of Coacalco (TESCo) — TecNM

Specialization in mobile application development.

Technology Industry

Platzi — Technology School

Specialized courses in web development, artificial intelligence and emerging technologies.

Programming Technician

CECyTEM — Tultepec

Specialized training in software development and development methodologies.

Certifications

Azure AZ-900: Microsoft Azure Fundamentals

2024

Microsoft

AzureCloudSecurity

Python for Everybody

2022

U. Michigan · Coursera

PythonSQL

Languages

SpanishNative
EnglishProfessional

Interests

Generative AIMLOpsDistributed systemsSpecialty coffeeTypography

Speaking

On stage

I give talks on AI and data science at universities — closing the loop between academic theory and what actually happens once a system meets production.

~200 attendees per talk

Where

EGADE Business School · Monterrey
UPVM Tultitlán
TESCo — TecNM

Topics

Generative AIRAG & MLOpsData science

Contact

Let's talk

I'm available for freelance projects, collaborations, and full-time opportunities. Tell me what you're building.

Send me a message

Reply within 24h · Mexico City, GMT-6

Emailhello@davidherrera.dev GitHub@david-heca LinkedInin/david-heca X@david_heca_

CV · ES CV · EN