Javier Garcia de Leaniz, Developer in Madrid, Spain
Javier is available for hire
Hire Javier

Javier Garcia de Leaniz

Verified Expert  in Engineering

Bio

Javier has 10+ years of experience in AI for both startups and Fortune 500 companies in diverse industries. A former CTO, founder, and EY Engineering Director, he specializes in building LLM-powered systems, advanced NLP pipelines, and Python back ends. His projects include enterprise chatbots, automating reviews of 2+ million government aid applications during COVID-19 in Spain (reducing manual effort by 95%), web scraping 1.5 million products from 1,000+ public sites, and many more.

Portfolio

Self-employed
Artificial Intelligence (AI), Python, Large Language Models (LLMs)...
Self-employed
Amazon Web Services (AWS), Flask, Natural Language Processing (NLP), React, SQL...
Smart Retrieval
Artificial Intelligence (AI), ChatGPT, OpenAI GPT-3 API, OpenAI GPT-4 API...

Experience

  • Python - 10 years
  • Artificial Intelligence (AI) - 7 years
  • Machine Learning - 7 years
  • Natural Language Processing (NLP) - 7 years
  • Open-source LLMs - 2 years
  • Retrieval-augmented Generation (RAG) - 2 years
  • OpenAI API - 2 years
  • LangChain - 2 years

Availability

Part-time

Preferred Environment

Azure, Amazon Web Services (AWS), Linux, Azure DevOps, Docker, OpenAI API, Python, FastAPI, PostgreSQL, LangChain, Spanish

The most amazing...

...product I've developed is an AI-powered tool that structures business data, documents, emails, and voice recordings and has processed millions of documents.

Work Experience

AI Engineer

2024 - 2025
Self-employed
  • Developed an AI-driven parsing system leveraging LLMs and NLP techniques to extract structured data from highly complex, hundred-page reports, reducing manual effort in risk assessment reviews by converting dense text into predefined JSON schemas.
  • Implemented advanced prompt engineering methods, few-shot learning, chain-of-thought prompting, prompt chaining, and structured outputs to handle domain-specific language.
  • Benchmarked open-source LLMs running locally and on AWS Bedrock (Phi 3.5, Llama 3.1, Mistral's) and OpenAI models on their API (GPT-4o and mini) using automated testing on a large, annotated dataset to ensure consistent performance and reliability.
  • Achieved 98% precision and 92% recall across 16 unique entity types, extracting over 1,500 data points from each document. These metrics were validated through automated comparisons to ground-truth annotations.
  • Reduced manual processing time by an estimated 95%, relying on a light-touch human-in-the-loop review. The solution is scalable to new document types as the parsing pipeline is easily configurable through simple options, accommodating any entity.
Technologies: Artificial Intelligence (AI), Python, Large Language Models (LLMs), Document Parsing, Pattern Recognition, Natural Language Processing (NLP), Llama, Mistral AI, Open-source LLMs, Amazon Bedrock, Amazon Web Services (AWS), Retrieval-augmented Generation (RAG), Text Classification, Transformers, LLM Evaluation, LLM Benchmarking, LangChain, Hugging Face, Function Calling & Tool Use in LLMs, AI Observability & Performance Monitoring, Bias Mitigation & Hallucination Reduction, Large Language Model Operations (LLMOps), Multistage LLM Chains, function calling, GitHub, Text Chunking Strategies, Data Processing

AI and Full-stack Engineer

2023 - 2025
Self-employed
  • Developed a web app that allows users to search for restaurants in natural language based on their characteristics. Created the full user interface, back end, AI modules, and CI/CD pipelines.
  • Designed a RAG pipeline, prompt-engineering a tool use module, and information retrieval with full-text search and LLMs such as GPT-3.5 and GPT-4o.
  • Designed and deployed the whole architecture using AWS stack such as Amazon EC2, Amazon RDS, Amazon S3, Elastic Load Balancing (ELB), etc.
  • Optimized the architecture, allowing to search over 3M restaurants in a fraction of a second on a monthly budget under $60.
Technologies: Amazon Web Services (AWS), Flask, Natural Language Processing (NLP), React, SQL, PostgreSQL, Python, OpenAI GPT-3 API, SQLAlchemy, OpenAI GPT-4 API, Information Retrieval, Prompt Engineering, Cognitive Computing, OpenAI, Custom Models, Web Scraping, Generative Pre-trained Transformer 3 (GPT-3), Data Science, Full-stack, Large Language Models (LLMs), Retrieval-augmented Generation (RAG), AI Prompts, APIs, Product Management, Supervised Learning, OpenAI API, Claude, Data Structures, AI Agents, PostgreSQL 9, Hugging Face, Full-stack Development, Function Calling & Tool Use in LLMs, AI Observability & Performance Monitoring, Bias Mitigation & Hallucination Reduction, Large Language Model Operations (LLMOps), Multistage LLM Chains, Vector Databases, function calling, Web Development, GitHub, Pydantic, Text Chunking Strategies, Amazon EC2, Amazon S3 (AWS S3), Scripting, Data Processing

CTO

2023 - 2025
Smart Retrieval
  • Led the technical strategy and product development of the company.
  • Deployed the platform to Azure using Azure DevOps CI/CD pipelines.
  • Developed a retrieval-augmented generation (RAG) pipeline to allow search in natural language over business documents such as financial statements, invoices, contracts, and more, leveraging OpenAI's API services.
  • Optimized the performance of the LLM-based functionalities by applying prompt engineering, fine-tuning LLMs, and using open-source LLMs such as Llama 3.
  • Developed an AI agent tasked with data validation and access to various tools such as mathematical operations, DB queries, and more.
Technologies: Artificial Intelligence (AI), ChatGPT, OpenAI GPT-3 API, OpenAI GPT-4 API, Azure, Python, OpenAI, B2B, Automation, Generative Pre-trained Transformer 3 (GPT-3), Data Science, Fine-tuning, Llama 3, Full-stack, Open-source LLMs, Large Language Models (LLMs), Retrieval-augmented Generation (RAG), LlamaIndex, Prompt Engineering, AI Prompts, Machine Learning Operations (MLOps), PDF, Llama, APIs, Product Management, Supervised Learning, Document Parsing, Pattern Recognition, OpenAI API, LangChain, Data Structures, Azure SQL Databases, FastAPI, Hugging Face, AI Chatbots, Chatbots, Function Calling & Tool Use in LLMs, AI Observability & Performance Monitoring, Bias Mitigation & Hallucination Reduction, Large Language Model Operations (LLMOps), Multistage LLM Chains, Vector Databases, function calling, AI Agents, Web Development, NSQL, GitHub, Pydantic, Text Chunking Strategies, Kubernetes, Technical Leadership, Leadership, Data Classification, Scripting, Data Processing, Django

AI Engineer

2023 - 2024
Explore My Store Pty Ltd
  • Created a web scraping process that obtained full details on 1.5 million products from 1400 eCommerce sites and optimized the process to detect product changes without requiring a full re-scrape. This optimization resulted in over 60% cost reduction.
  • Developed an ETL to load and transform the scraped data into an Azure Cosmos DB, enriching the data using LLMs to determine product categories, create embeddings to allow vector search, and more.
  • Configured and optimized an Azure Search service, working with the product team to integrate it with the web application. This included search functionalities such as full-text search, vector search, facets, filters, spelling correction, etc.
  • Developed a web scraping process to obtain full details on eCommerce stores, such as the company logo, about us section, payment methods accepted, social media links, and more.
  • Developed a process to determine if a website is down, detecting edge cases such as deactivated Shopify stores, domains for sale, websites under maintenance, and more.
Technologies: Artificial Intelligence (AI), Azure, Azure Cognitive Services, OpenAI GPT-4 API, ChatGPT, Generative Artificial Intelligence (GenAI), Azure Search, Python, Azure Cosmos DB, Web Scraping, Large Language Models (LLMs), Prompt Engineering, Data Science, Full-stack, AI Prompts, Large Data Sets, eCommerce, Artificial Neural Networks (ANN), Supervised Learning, OpenAI API, Data Structures, Azure SQL Databases, Hugging Face, Data Pipelines, NSQL, GitHub, Scraping, Data Classification, Data Processing

Engineering Director

2016 - 2023
EY
  • Led a multidisciplinary product team of 50+ team members building AI-driven products with a focus on NLP and large language models (LLMs).
  • Integrated 5+ redundant AI solutions on a single company-wide platform as part of a global rationalization initiative, choosing architectural components, AI models, and user experience based on client requirements and internal benchmarks.
  • Developed, trained, and evolved multiple models for different functionalities, including layout detection, document classification, named entity recognition, and question-answering and section ranking models.
Technologies: Python, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Machine Learning, Docker, Artificial Intelligence (AI), Azure, ChatGPT, OpenAI GPT-4 API, OpenAI GPT-3 API, Agile Software Development, Azure Cognitive Services, OpenAI, Data Scraping, Architecture, API Integration, B2B, Automation, Integration, Consulting, Data Science, Full-stack, Large Data Sets, Machine Learning Operations (MLOps), PDF, APIs, Databricks, GPU Computing, Speech Recognition, Speech to Text, Graphics Processing Unit (GPU), Speech Analytics, Deep Learning, PyTorch, Electronic Health Records (EHR), Machine Learning Algorithms, Product Management, Data Analysis, Supervised Learning, Document Parsing, Pattern Recognition, OpenAI API, LangChain, Speech to Text AI, Data Structures, Azure SQL Databases, Transformers, Convolutional Neural Networks (CNNs), PostgreSQL 9, Hugging Face, AI Chatbots, Chatbots, Data Pipelines, Function Calling & Tool Use in LLMs, AI Observability & Performance Monitoring, Bias Mitigation & Hallucination Reduction, Large Language Model Operations (LLMOps), Multistage LLM Chains, Vector Databases, function calling, Regression Modeling, Gradient Boosting, Logistic Regression, Random Forest Regression, Linear Regression, Decision Tree Regression, GitHub, Text Chunking Strategies, Recurrent Neural Networks (RNNs), Kubernetes, Robotic Process Automation (RPA), Technical Leadership, Leadership, Data Classification, Data Analytics, Decision Trees, Scripting, Data Processing, Django

Data Engineer

2014 - 2016
Accenture
  • Designed and developed risk assessment processes for multichannel applications (smartphone app, web, ATM, bank branch) of a Spanish international bank.
  • Developed data pipelines for the risk assessment process of credit cards and online personal loans.
  • Developed SQL queries to analyze risk customer data indicators.
Technologies: SQL, Scrum, Data Engineering, Data Analysis, Data Structures, Data Pipelines, Data Analytics, Data Processing

Experience

Social Media Restaurant Review Videos Analysis

I developed a back-end application that inputs a social media link (such as a TikTok video) and analyzes its content to identify restaurant reviews.

The pipeline works by downloading the media, analyzing photos, creating a transcript of videos, and analyzing the video images to find out key information about restaurants, such as locations, prices, dishes, and more.

This data is stored in MongoDB so a FastAPI back end can search for the restaurant's data.

I built this system so I could create a wish list of restaurants based on my preferences that I can search using natural language queries, such as "a typical Spanish restaurant with authentic paella."

Tax Relief Application Eligibility

I built and trained NLP models to extract key data points from various documents such as invoices, mortgage payments, paychecks, and more.

The goal was to increase the efficiency of the application process for a tax relief program offered by the government due to COVID-19, which received millions of requests.

I developed a pipeline consisting of handwritten text detection, layout detection techniques, classification (to detect the document type), and NER (named-entity recognition) models to accurately extract the relevant information from the documents. I also developed the confidence module, which prioritized the manual review of applications based on the confidence of business rules and models.

This project allowed the government to process the sudden millions of requests in time for businesses to benefit from the subsidies, which would have been impossible if done manually.

Insurance Claim Payment Automation

I trained and developed NLP models to identify, extract, and structure data from veterinary invoices to allow for the reimbursement of animal health insurance claims.

I developed a pipeline consisting of OCR, layout detection techniques, and named-entity recognition (NER) models to accurately extract the relevant information from the invoices. I also built the validation module to identify and validate medical diagnoses against the policyholder coverage.
Finally, I developed the extraction confidence methodology to help determine claims reimbursements to be processed automatically or reviewed by a human, depending on the different models' confidence and business rules.

Mortgage Contract Audit Automation

I trained and developed NLP models to identify key data points from mortgage contracts, allowing automatic audit and validation of data in actual contracts versus the ERP.

I trained a model to classify between main contracts and their annexes, extensions, and modifications using TF-IDF features to train a classifier. I also developed the validation module to disambiguate and match contracts and DB rows and perform the comparison to highlight differences.

Invoice Validation Automation

I developed an AI-driven system to automate the analysis of financial documents in construction processes, targeting the high volume of invoices, purchase orders, and goods received notes. This project aimed to detect mismatches and inconsistencies to prevent financial losses.

The project's pipeline started with OCR technology to accurately extract text from scanned documents. I used named-entity recognition (NER) models to identify and categorize key data points within these texts, such as vendor names, dates, and amounts. An important part of the project was the development of classification models to accurately detect and categorize different document types, automatically detecting their relevant data points.

Additionally, I implemented fuzzy matching algorithms to link items listed in invoices with corresponding entries in purchase orders. This approach was key in identifying mismatches and inconsistencies.

Education

2007 - 2013

Master's Degree in Industrial Engineering

Universidad Pontificia Comillas - Madrid, Spain

Certifications

DECEMBER 2018 - PRESENT

Natural Language Processing Nanodegree

Udacity

DECEMBER 2017 - PRESENT

Machine Learning Engineer Nanodegree

Udacity

Skills

Libraries/APIs

OpenAI API, SpaCy, Scikit-learn, Pandas, Azure Cognitive Services, PyTorch, Pydantic, Keras, React, SQLAlchemy

Tools

ChatGPT, AI Prompts, GitHub, Named-entity Recognition (NER), Azure Search, Whisper

Languages

Python, SQL

Frameworks

Flask, LlamaIndex, Django

Paradigms

Automation, Scrum, Agile Software Development, B2B, Azure DevOps

Platforms

Azure, Docker, Amazon Web Services (AWS), Visual Studio Code (VS Code), MacOS, Linux, Windows, Databricks, Amazon EC2, Kubernetes

Storage

Azure SQL Databases, Data Pipelines, NSQL, PostgreSQL, Azure Cosmos DB, MongoDB, Amazon S3 (AWS S3)

Other

Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Artificial Intelligence (AI), OpenAI GPT-4 API, OpenAI GPT-3 API, Prompt Engineering, OpenAI, Large Language Models (LLMs), Retrieval-augmented Generation (RAG), Document Parsing, Pattern Recognition, Function Calling & Tool Use in LLMs, function calling, Spanish, Machine Learning, Deep Learning, Optical Character Recognition (OCR), Information Retrieval, Data Scraping, Web Scraping, Generative Artificial Intelligence (GenAI), Data Science, Fine-tuning, Open-source LLMs, Large Data Sets, Machine Learning Operations (MLOps), PDF, Llama, APIs, Artificial Neural Networks (ANN), Machine Learning Algorithms, Product Management, Supervised Learning, LangChain, Data Structures, AI Agents, FastAPI, PostgreSQL 9, Hugging Face, AI Chatbots, Chatbots, AI Observability & Performance Monitoring, Bias Mitigation & Hallucination Reduction, Large Language Model Operations (LLMOps), Multistage LLM Chains, Vector Databases, Text Chunking Strategies, Scraping, Technical Leadership, Leadership, Data Classification, Decision Trees, Scripting, Data Processing, Computer Vision, Handwriting Recognition, Text Classification, Tf-idf, Cognitive Computing, Custom Models, Architecture, API Integration, Integration, Consulting, Generative Pre-trained Transformer 3 (GPT-3), Llama 3, Full-stack, Data Engineering, GPU Computing, Speech Recognition, Speech to Text, Graphics Processing Unit (GPU), Speech Analytics, eCommerce, Electronic Health Records (EHR), Data Analysis, Claude, Speech to Text AI, Transformers, Convolutional Neural Networks (CNNs), Mistral AI, Amazon Bedrock, LLM Evaluation, LLM Benchmarking, Full-stack Development, Web Development, Software Development, Regression Modeling, Gradient Boosting, Logistic Regression, Random Forest Regression, Linear Regression, Decision Tree Regression, Recurrent Neural Networks (RNNs), Developing AI Models locally, Robotic Process Automation (RPA), Data Analytics

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