Martí Jordà Roca

Software Engineer @ NeuralTrust

About

Top projects that I have been involved in: 1. 🐱 Machine Learning Platform for Wallapop: - Designed and built a robust machine learning platform at Wallapop. - Using tools such as Sagemaker, Cloudformation, Flyte, Kubernetes, and Feast. - Simplified the process of deploying machine learning models into production. 2. 🐹 Production ML Models: - Successfully deployed several ML models, including: - Marketing churn prediction - Category classifier - Weight suggestion model - Utilized Flyte, Sagemaker Endpoints, and Grafana for efficient monitoring. 3. 🐼 Personalization Model: - Maintain and improve an ALS collaborative filtering model with implicit feedback. - Using Spark, Airflow, and MLflow for seamless model management. 4. 🐶 Google Analytics Streaming Data Processing: - Processed clickstream data from Google Analytics. - Stored the processed data in Delta Lake using Spark Structured Streaming. 5. 🐷 Custom Python ML Model Framework: - Created a flexible framework for saving, loading, tracking, and deploying custom Python ML models. - Integrated MLflow, AWS Sagemaker, and Python for streamlined workflows. About Me: Born in the small town of Vic and currently living in Barcelona. My passion lies in the realms of Big Data, Machine Learning, and Deep Learning, particularly in taking these models to production. I thrive on learning about cutting-edge machine learning technologies and architectural advancements. Beyond my professional pursuits, I’m an energetic and proactive individual. Whether it’s a game of handball, a session at the gym, or diving into new knowledge, I’m always seeking ways to enhance my skills and contribute effectively. Feel free to connect—I’m eager to engage with fellow enthusiasts and collaborate on exciting projects! 🚀

Country

Spain

City

Vic

Industry

Computer Software

Skill

Large Language Models (LLM), Google Cloud Platform (GCP), Apache Kafka, Flyte, Kubernetes, Feast, Huggingface, Amazon EKS, Databricks, Architecture, Apache Spark, Amazon Web Services (AWS), Python (Programming Language), Computer Science, Big Data, Machine Learning, Deep Learning, Java, C++, SQL

Experience

Software Engineer

NeuralTrust

2024-10 - Present · 2 yrs

Barcelona, Catalonia, Spain

Building the platform to secure, analyze and understand GenAI.

Machine Learning Engineer

Wallapop

2022-7 - 2024-10 · 2 yrs 4 mos

Barcelona, Catalonia, Spain

Designing and building the machine learning platform for Wallapop to simplify and facilitate taking into production machine learning models. Using Sagemaker, Cloudformation, Flyte, Feast, k8s and Grafana. Other top projects that I have been involved in: - Define standard ML project Software architecture. - Taking into production models for churn, item category classification, weight suggester. - Designing architecture for item embeddings personalization model. Using Open Search, Sagemaker Batch Transform Jobs. - Web app accessible for the whole company to show our inhouse ML models. Using Streamlit and ECS. - Reduced cost of Sentence Transformers online inference model up to a 90%, compiling it with AWS neuron. - MVP for developing ML pipelines with Flyte using EKS. - Automatic Background Removal model MVP.

Data Scientist, personalization

MANGO

2020-6 - 2022-7 · 2 yrs 2 mos

Barcelona Area, Spain

Projects involved: - Build a Qlik dashboard for Google clickstream data analysis and its data pipeline. Using Qlik Sense, Airflow, Pyspark, S3 and AWS Glue. - Build a Qlik dashboard for sales analysis and its data pipeline. Using Qlik Sense, Airflow, Pyspark and S3. - Developed a framework for saving, loading, tracking and deploying any custom python machine learning models to AWS Lambda or Sagemaker. - Design and implementation of a real-time inference model for nearest-neighbors clothes.

Data Engineer

MANGO

2019-6 - 2020-6 · 1 yr 1 mo

Barcelona, Catalonia, Spain

Projects: - Realtime clickstream recommendation system using scala with spark, AWS Kinesis, DynamoDB, AWS Lambda and API Gateway and S3. - Develop a ETL process for the company Data Lake which writes as Delta any data file written in a folder of the Data Lake. Using AWS Lambda, PySpark and S3. - Orchestration of a data pipeline for a machine learning model that finds similar clothes to each other. Using Airflow, PySpark, Scala and DynamoDB. - Serverless webpage POC using Angular, Typescript as frontend and AWS Lambda, DynamoDB and APY Gateway as backend.