Patricio is available for hire
Hire PatricioPatricio Requena
Verified Expert in Engineering
Python Developer
Quito, Pichincha, Ecuador
Toptal member since October 21, 2026
Bio
Patricio is an AI and machine learning engineer with over four years of experience building production systems for clients including Rappi, Factored AI, and Thoughtworks. His primary expertise is in Python, FastAPI, and Airflow for industries such as retail, CPG, and airlines. He achieved a 95% rejection rate for off-topic and harmful queries by designing multi-layered guardrails for production agentic systems.
Portfolio
Factored AI
Python, FastAPI, Amazon Web Services (AWS), LangGraph, PydanticAI, PostgreSQL...
Thoughtworks
LangChain, LangSmith, FastAPI, Pydantic, Google Cloud Platform (GCP), Vertex AI
Rappi
FastAPI, PostgreSQL, Large Language Models (LLMs), Pgvector
Experience
- Python - 5 years
- Machine Learning - 4 years
- SQL - 4 years
- Recommendation Systems - 3 years
- FastAPI - 3 years
- Artificial Intelligence (AI) - 3 years
- AI Agents - 2 years
- LangGraph - 2 years
Work Experience
Machine Learning Engineer
2026 - 2026
Factored AI
- Designed, built, and deployed end-to-end GenAI systems for the financial and healthcare industries, owning the full lifecycle from exploratory data analysis of internal corpora to production deployment on AWS.
- Built an agentic RAG system for financial document analysis with LangGraph, integrating 3rd-party financial APIs to supplement retrieved context with live market data.
- Orchestrated a multi-agent healthcare system with PydanticAI, integrating external medical APIs for terminology lookup and enforcing data protection controls to prevent patient information from being exposed to 3rd-party providers.
- Developed PDF parsing, chunking, and embedding pipelines ingesting documents from AWS S3 into PostgreSQL with PGVector, applying hybrid search, metadata filtering, and query decomposition to improve retrieval precision.
- Implemented multi-layered guardrails combining program logic and LLM-as-a-Judge validation, achieving a 95% rejection rate of off-topic and harmful queries across the evaluation dataset.
Technologies: Python, FastAPI, Amazon Web Services (AWS), LangGraph, PydanticAI, PostgreSQL, Pgvector
Data Scientist AI Consultant
2025 - 2026
Thoughtworks
- Designed and developed LLM based solutions to automate operational processes at one of the largest airlines in LATAM using LangChain to orchestrate and parse the GenAI process together with LangSmith to monitor the workflow performance.
- Reduced by 40% the human-hours invested in repetitive tasks with an AI automation process.
- Built data and AI pipelines using Google VertexAI to orchestrate the entire workflow for multiple products and GCP Storage for file management.
- Developed FastAPI microservices following best practices such as data validation using Pydantic to expose LLM processes and being integrated with other teams.
Technologies: LangChain, LangSmith, FastAPI, Pydantic, Google Cloud Platform (GCP), Vertex AI
GenAI Machine Learning Engineer
2023 - 2025
Rappi
- Participated in the development of a GenAI platform to improve business processes and democratize the usage and understanding of different tools based on AI solutions for technical and non-technical users at a giant LATAM-based food delivery.
- Deployed microservices with FastAPI and PostgreSQL to enhance projects using GenAI, solving key business challenges.
- Performed LLM fine-tuning for customer support-related tasks with an AI Buddy, improving by 20% the first response time between agents and customers.
- Worked with RAG systems for easy access to internal knowledge using vector databases such as PGVector and ChromaDB.
Technologies: FastAPI, PostgreSQL, Large Language Models (LLMs), Pgvector
Data Scientist
2022 - 2023
Rappi
- Worked on recommendation projects in the CPG team for a giant LATAM-based food delivery company.
- Increased conversion rates by 3% by improving the current recommender system variables.
- Analyzed model performance with online metrics by running A/B testing experiments with multiple user groups using pandas for data management and NumPy for matrix operations.
- Built data pipelines using Apache Airflow to orchestrate data tasks from data cleaning and model training.
- Used SQL to retrieve necessary data from different sources such as Snowflake warehouse and performed data analysis through Jupyter notebooks together with matplotlib and seaborn to generate understandable visualizations.
Technologies: Pandas, NumPy, Apache Airflow, SQL, Snowflake, Matplotlib, Seaborn, A/B Testing
Data Scientist
2021 - 2022
Self-employed
- Delivered end-to-end data solutions for clients in the education and retail sectors, utilizing Python (Pandas, NumPy), SQL, and Tableau.
- Designed interactive Tableau dashboards to monitor financial KPIs, replacing manual Excel workflows.
- Engineered Scikit-learn classification models for a retail inventory workflow, improving F1-score by 10% on imbalanced datasets.
- Optimized feature selection to ensure consistent performance across diverse product categories.
Technologies: Python, Pandas, NumPy, SQL, Tableau, Scikit-learn
Project History
Computer Vision-based Autonomous Navigation for UAV Swarms | Springer Research
https://doi.org/10.1007/978-3-031-83210-9_11I co-authored a peer-reviewed research paper published by Springer Nature on autonomous UAV navigation and swarm coordination using computer vision. I led the computer vision development, implementing object detection and ArUco marker-based tracking to support visual identification and navigation.
I also designed the communication architecture using a centralized coordination system, in which a central computer served as a gateway for exchanging information and coordinating tasks across multiple drones. My contributions focused on integrating visual perception, communication, and multi-UAV coordination into a unified robotics system to address challenges in autonomous navigation and collaborative task execution.
The research was presented at ARTIIS 2024 and published by Springer in 2025.
I also designed the communication architecture using a centralized coordination system, in which a central computer served as a gateway for exchanging information and coordinating tasks across multiple drones. My contributions focused on integrating visual perception, communication, and multi-UAV coordination into a unified robotics system to address challenges in autonomous navigation and collaborative task execution.
The research was presented at ARTIIS 2024 and published by Springer in 2025.
Education
2025 - 2026
Bachelor's Degree in Artificial Intelligence
International University of Ecuador - Quito, Ecuador
2020 - 2024
Bachelor's Degree in Electronics and Automation Engineering
Salesian Polytechnic University - Quito, Ecuador
Certifications
SEPTEMBER 2026 - SEPTEMBER 2027
Claude Certified Architect
Anthropic
AUGUST 2022 - PRESENT
Astronomer Certification for Apache Airflow 2
Astronomer
Skills
Libraries/APIs
Pydantic, Pandas, NumPy, Matplotlib, Scikit-learn, PyTorch, Claude API, OpenCV
Tools
Apache Airflow, Seaborn, Tableau, Git, Claude, Claude Code
Languages
Python, SQL, Snowflake
Frameworks
LangGraph
Platforms
Docker, Amazon Web Services (AWS), LangSmith, Google Cloud Platform (GCP), Vertex AI
Storage
PostgreSQL, Data Pipelines
Other
Artificial Intelligence (AI), Machine Learning, FastAPI, Software, AI Agents, Natural Language Processing, Recommendation Systems, Information Retrieval, PydanticAI, Pgvector, LangChain, Large Language Models (LLMs), GCP, Google BigQuery, Distributed Systems, Electronics, Robotics, Computer Vision, Agentic AI, Directed Acrylic Graphs (DAG), Orchestration, Object Detection, Swarm Robotics, Autonomous Navigation, A/B Testing
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring