Baja California, Mexico
Check out my personal website for a complete summary of my skills and experience: https://erickmarquez.com
Worked across a large Python monorepo powering AI-driven features for a YouTube creator tools platform. Built and maintained production microservices deployed on Kubernetes, with a focus on ML infrastructure, LLM integrations, and backend engineering. - LLM Gateway: Core contributor to a unified proxy service that routes requests across multiple model providers, handling request transformation, streaming, token tracking, and deployment management. - Core API & Background Workers: Key contributor to the central FastAPI backend serving 30+ AI-powered feature endpoints — including thumbnail scoring, title generation, video understanding, content optimization, and script generation — backed by PostgreSQL, Redis, Celery task queues, and DynamoDB. - Keyword Scoring & Recommendation Engine: Took ownership of an existing service, implementing all MongoDB and DynamoDB data access layers, scoring algorithms, trend analysis, and search pipelines. - ML-Powered Image Processing Service: Led the initial implementation of an autocropping service, optimizing it for horizontal pod autoscaling on Kubernetes to handle bursty production workloads efficiently. - Model Evaluation Tooling: Designed and built an internal CLI tool for evaluating LLM prompts and model performance using code markers, enabling the team to systematically compare model outputs and track prompt quality over time. - Developer Experience & AI-Assisted Workflows: Drove adoption of AI-assisted development across the team — authored coding standards, agent configurations, and MCP server integrations, helping establish conventions for the entire monorepo. Tech: Python 3.12, FastAPI, Celery, PostgreSQL, MongoDB, DynamoDB, Redis, Kafka, SQLAlchemy, Docker, Kubernetes, Helm, ArgoCD, GitHub Actions, Prometheus.
Owned and built the core data platform powering analytics for a creator intelligence product serving thousands of YouTubeusers. Designed and maintained Airflow DAG categories orchestrating data ingestion, transformation, and delivery across a modern AWS data lake (S3, Athena, Glue, Iceberg). Built both heavy-duty PySpark pipelines on EMR for large-scale processing and lightweight Pandas-based ETL for smaller workloads. Worked across the full infrastructure stack — containerized services with Docker, deployed them on EKS via Helm and ArgoCD, and managed cloud resources with Terraform. Integrated data from YouTube, Kafka streams, MongoDB, PostgreSQL, and third-party platforms into a centralized data lake with landing, staging, and production layers. Contributed to ML-adjacent workloads including semantic embeddings served through OpenSearch, topic prediction with sentence-transformers, and AI-powered coaching analysis using OpenAI. Set up monitoring with Prometheus exporters, enforced code quality through comprehensive CI/CD pipelines in GitHub Actions, and managed secrets. Stack: Python, Apache Airflow, PySpark, SQL, AWS (S3, EMR, EKS, Athena, Glue, Lambda, RDS), Kafka, Terraform, Docker, Kubernetes, Helm, ArgoCD, OpenSearch, MongoDB, PostgreSQL, GitHub Actions.
primiciaens.com is a website I created to serve as a hub for all the different news happening in my local community. It crawls all the different newspaper websites and serves them all in a friendly UI that can be installed as a native app, thanks to the power of PWAs. The technologies powering the project are: Dokku, Preact, Python and Postgresql.
Data engineer working with: * Apache AIrflow * Amazon S3 * Amazon Athena * AWS Serverless architecture services * AWS EMR * PySpark * Python * Pandas * MySQL * Docker * JavaScript/Vue.js
My current position in the company is that of a full stack developer. Working primarily with our main product, a WMS (Warehouse Management Software). The Symfony php framework serves the core functionality and we make full use of all the available features, migrations, entities, services, twig, custom commands, to name a few. While most of the legacy frontend logic is handled by jquery and libraries, I've worked on introducing webpack to the stack. Working in a mechanism to make use of the translations and routing from symfony inside our new view components. This was done by developing a webpack loader and plugin to parse the source code while it is being built, to find all the instances of the translation and route functions. Webpack lets us make use of React, Typescript and Redux for any new component we need. The rest of the stack is composed of Mysql as the database engine, Mercurial for version control and Ubuntu as my main operating system. The quality of the software is taken care by a full set of different tests like behat, unit tests and functional tests. Docker is also thrown into the mix, we use it constantly to test the software we're reviewing, this helps us setup a clean environment that is not contaminated by local changes we could be working in our local instances.
I work here in my spare time from my daily work. My role is that of a software engineer, database manager and administrator, and consultor. My first experience here was developing a map section for the wordpress site, designing and developing the database, creating the front-end view via the Google Maps API, customising the aspect and design of the map and markers. This part of the website was then managed by an admin system that I created inside of the wordpress admin ecosystem, to manage local business, categories, images, etc. A calendar section was then created to handle events, as well as the corresponding admin backend section. I was then given the task of creating a RESTful API to communicate with their Android and iOS mobile apps. This was done with the help of php and slim framework. As the system grew bigger and bigger, I had to separate the wordpress admin system from our own management system to have more freedom on how to handle some of the tasks. I then used the Laravel framework to re-write the whole admin system and redesign the database. Between many upgrades, one of the primary ones was the option to resize the uploaded images to fit the different devices served by our API. I'm currently developing another wordpress site and a wordpress theme. Skills: - php development - javascript and jquery development - Google Maps API V3 Javascript usage - Database management and administration - VPN/Server instance creation and setup - Wordpress developing and management - Sublime Text - phpstorm - vim/macvim - Laravel Framework development