Big Things 2019

Joaquín Montesinos
Jeff Tech
Published in
4 min readDec 18, 2019

On November 20th & 21st, Big Things 2019 8th edition event was organized by Paradigma Digital in Madrid. This event was born in 2012 with the aim of raising awareness of Big Data among the business community in Spain.This year the event dealt with Big Data, Artificial Intelligence, Machine Learning, IoT, Blockchain and Quantum Computing with some of the best speakers related to data (https://www.bigthingsconference.com/2019/speakers) from the top world’s companies.

Big Things Conference: “It’s a great conference, high-quality speakers, a really good mix of topics”

One of Jeff’s goals is to boost their employees knowledge so it was a great opportunity for Aitor Miota, Joaquín Montesinos and Juanjo Milla, representing Jeff, to keep up to date with the latest technologies and architectures around data and how we could apply these new technologies on Jeff.

The event had 5 tracks in parallel, 2 techniques and 3 business, in addition to the stands where they made product demonstrations. It was difficult to decide which of the tracks to attend at all times, but we played an intelligent strategy: the team has two very strong profiles such as the data scientist and the data engineer, so we divided the tracks to get more out of it. If interested, this is a summary of the tracks that took place and the youtube channel: https://www.bigthingsconference.com/2019/schedule/, https://www.youtube.com/user/BigDataSpain

Jeff’s Data Team at Big Things

TALKS

#onejourneyWiser (Carlos Herrera):

In this talk they explained the phases they have followed since Cabify to create their machine learning platform (Lykeion). They also explained the flows and tools they used for the deployment, publish and training of their models.

It is very important for Jeff’s data team to serve models with agility, in addition to having absolute control to test, validate and version these models. Talks like this provide a vision of how to prepare the structure for when Jeff deploy a high number of models in production and to allow any member of the team to serve their models quickly. As an example, our location model allows us to know the best places to configure our centers. Currently, the deployment method is based in a jenkins that build a docker image, directed by kubernetes, where is defined the api that serve the model for the company.

Solving Natural Language Problems with scarce data (Álvaro Barbero Jiménez):

This talk taught us different techniques to address text analysis problems and the evolution of an NPL algorithm from a basic level to a more advanced algorithm, comparing success coefficients and techniques. One of the most interesting things they showed us was BERT (Bidirectional Encoder Representations from Transformers), a pre-trained NLP model developed by Google. Bert makes use of an attention mechanism that learns contextual relations between words in a text.

At Jeff we have several algorithms that require NLP for our partners product improvement. For instance, we’ve been analyzing the creation product process to categorize products created by our partners, trying to match that product with our global products. That algorithms will provide us information about how partners create their own products (and how often) or if they have some needs related to new products creation.

The future was this: serverless face recognition in real-time video (Javier Ramírez):

In this talk an example of using the Amazon Rekognition tool was explained, in which Javier Ramírez, 13, was identified by face recognition using an image of him in 2019.

Multiple benefits from using cameras at Jeff: from the videos we are actually recording to the project we expect to have during 2020 about getting insights from monitoring camera recording.

AWS DeepRacer:

As an incentive to the entire event, an autonomous car racing competition was held through the use of AWS DeepRacer. For this competition, you could bring a trained home model and compete against the time of the other attendees to the event. It was a fun initiative to start learning Reinforcement Learning (RL).

SUMMARY

Big Things 2019 8th edition was an incredible experience for Data Team, where we met incredibly people, learned a lot and shared Jeff to other companies.

“With a stage magician as the MC and a rock band on stage, Big Things 2019 was a lot of fun!”

Aitor Miota, Data Analyst at Jeff & Joaquín Montesinos, Business Intelligence Analyst

--

--