Skip to content

AI & Automation Engineer

Carlos Valencia

Integrations · AI Agents · ERP

Mechatronics Engineer focused on AI, automation and system integration. I turn manual processes and ambiguous requirements into reliable solutions, connected to the platforms a business already runs on.

Guadalajara, MexicoSpanish (native) · English (B2)

3+

years building production systems

90+

custom ERP modules in the codebase

285

commits as 2nd contributor on the ERP monorepo

10

shipped projects documented as case studies

Selected work

Source code for employer and client projects is proprietary and not published. These are write-ups: the problem, what I built, the stack, and my actual role — with commit counts where they are verifiable.

Odoo ERP Platform

2025—2026

ERP · Backend

Custom Odoo 18 modules running production, sales, logistics and billing for an ice and cold-chain manufacturer.

285commits · 2nd of 10 contributors
  • Odoo 18
  • Python
  • PostgreSQL
  • OWL
  • n8n
  • WhatsApp Business

KoldPlant — Production Telemetry

2026

IoT · Data Architecture

Plant telemetry ingested into the ERP, reconciled against human capture, and surfaced by role.

  • IoT telemetry
  • n8n
  • Odoo 18
  • PostgreSQL
  • SQL views
  • Metabase

ERP-Integrated Operations App

2026

PWA · Full-stack

A React PWA to run and record plant, sales and delivery operations live against the ERP.

187commits · 2nd of 6 contributors
  • React 18
  • Vite
  • Tailwind CSS
  • Odoo REST
  • PostgreSQL
  • Vercel

KoldHome — D2C Ordering PWA

2026

E-commerce · Payments

Consumer storefront with passwordless login and card payment resolved through the ERP.

  • Next.js
  • TypeScript
  • Odoo REST
  • Stripe
  • Vercel
  • n8n

B2B Ordering PWA

2026

E-commerce · Frontend

Self-service ordering for retailers, against a catalog derived live from the ERP category tree.

  • Next.js
  • TypeScript
  • Tailwind CSS
  • Odoo REST
  • Vercel

Inspections & Reporting on Power Platform

2025

Microsoft 365 · Low-code

A Power Apps canvas app, SharePoint data model and Power Automate flows that turn field inspections into signed PDF reports.

~396rows rendered per consolidated report
  • Power Apps
  • Power Automate
  • SharePoint
  • Word/PDF
  • Microsoft 365

Conversational Commerce — WhatsApp to ERP

2025—2026

AI Agents · Commerce

A chatbot that takes an order through WhatsApp and drives it all the way to a warehouse pick — writing to the ERP, not just answering questions.

43commits · 2nd of 5 on the bot controller
  • Botpress
  • WhatsApp Business API
  • Meta API
  • n8n
  • OpenAI
  • Odoo
  • Python

AI Agents & Business Automations

2024—2026

AI · Automation

AI agents and chatbots wired into the systems where the business actually runs.

  • OpenAI
  • Claude
  • Botpress
  • n8n
  • Power Automate
  • Python

Ofertin — WhatsApp Offers Channel

2025

Own product

An API that runs a WhatsApp promotions and referrals channel. Built from scratch.

  • WhatsApp Business API
  • REST APIs
  • Webhooks
  • Python
  • JavaScript

Marketplace Automation

2024—2025

Integration · Automation

Listing and catalog data kept current across marketplaces, without manual checks.

  • Mercado Libre APIs
  • Web scraping
  • n8n
  • Google Sheets
  • Python/JS

How I build

The same three ideas show up in almost everything above. They are what separate a demo from a system that survives contact with an operation.

Telemetry → truth: the KoldPlant intake path

Machine data never becomes a business number in one hop. It lands as evidence, gets reconciled against human capture, and only then drives a decision.

Telemetry → truth: the KoldPlant intake pathPlant equipment (telemetry) → n8n (ingest + map) → Intake tray (idempotent · evidence) → Reconciliation (vs. human capture) → Validated production (auditable state) → Surfaces by role (PWA · BI · finance)01Plant equipmenttelemetry02n8ningest + map03Intake trayidempotent · evidence04Reconciliationvs. human capture05Validated productionauditable state06Surfaces by rolePWA · BI · finance
intake pipeline · schematic
  • 00.01telemetry device → intake tray
  • 00.02intake idempotency key matched · duplicate dropped
  • 00.03reconcile machine output vs. shift capture
  • 00.04validate variance within tolerance
  • 00.05post production validated · costed

01

Idempotency is not optional

Field devices retry on bad connectivity. Every write path I build — packing capture, WhatsApp sends, webhook intake — is guarded against double-submission, because in the field a retry is the normal case, not the edge case.

02

One source of truth, several surfaces

An operator, a plant manager and finance need different detail and different permissions over the same data. I build one logic and one truth, then present it per role — rather than letting a dashboard quietly become the system of record.

03

Correctness at the source

Timezones, tax rounding, company scoping. A daily close that computes 'today' in UTC will be wrong every single day until it is fixed where the data is written, not patched in the report.

Stack

What I reach for. Depth varies — the ERP and automation side is where I have the most production mileage.

AI
OpenAIClaudeBotpressPrompt engineeringConversational agents
Automation
n8nPower AutomateWebhooks
ERP
Odoo 18PythonPostgreSQLOWL
Web
Next.jsReactTypeScriptTailwind CSSVite
Integration
REST APIsWhatsApp BusinessMeta APIStripeMercado Libre
Microsoft 365
Power AppsPower AutomateSharePoint
Other
GitMQTT / IoTMetabaseVercel

Contact

Open to AI engineering, automation and integration roles — remote or in Guadalajara.