Axel Sirota
Founder @ Data Trainers
Buenos Aires, Argentina
Actions
I'm the founder of Data Trainers LLC and a PhD candidate in Machine Learning at the University of Buenos Aires, focusing on NLP, LLMs, and Generative AI. I enjoy creating engaging courses for Pluralsight, O'Reilly Media, and LinkedIn, and love sharing my knowledge through global corporate trainings.
Links
Topics
Unleashing Generative AI in Real-Time: Transforming Customer Feedback with BERT on Apache Pulsar
Dive into the realm of real-time sentiment analysis with me! We’ll transform customer experiences by integrating a BERT sentiment model into Apache Pulsar as a Pulsar Function. This isn’t just analysis; it's about revolutionizing how we understand and react to customer feedback instantly with the power of Generative AI
Let's build an LLM from scratch!
LLMs (large language models) are huge models based on the Transformer architecture introduced in Vaswani's paper. They have a lot of mystique and this 2023 they exploded due to the use that could be given to them thanks to ChatGPT, DALL-E and friends.
In this workshop I propose that, with patience, we see the step by step of building an LLM of 0 in Keras and with HuggingFace. We are going to train it and we are going to solve Q&A with context, that is, given a question and a context, what is the answer. Along the way we will learn about tokenization, batching, finetuning, transfer learning, and much more!
Creemos una LLM desde 0!
Los LLM ( modelos de lenguage grandes) son unos modelos enormes basados en la arquitectura del Transformer introducido en el paper de Vaswani. Tienen mucha mistica y este 2023 explotaron por el uso que se le pudo dar gracias a ChatGPT, DALL-E y allegados.
En este taller propongo que, con paciencia, veamos el paso a paso de construir en Keras y con HuggingFace una LLM de 0, vamos a entrenarla y vamos resolver Q&A con contexto, o sea dada una pregunta y un contexto, cual es la respuesta. En el camino vamos a aprender de tokenizacion, batching, finetuning, transfer learning, y mucho mas!
Concurrency in Python from the ground up
Many Python applications nowadays need to handle concurrent users. However, over the years, the concurrency paradigm in Python migrated from threads or processes to asyncio.
This session will review how to solve the same problem with coroutines, multithreading, multiprocessing and asyncio to remark on differences, pros, cons and lessons learned!
Interactive Analytics on Pulsar with Pulsar SQL
Suppose you want to know analytics on your Pulsar topics, or you want to debug those hard corner cases that fail to be sent, or even you want to monitor your Pulsar deployment: how do you do it?
A tool exists to do this and more: Pulsar SQL. Since the 2.2.0 release, Pulsar SQL provides an abstraction layer to run any SQL query we may want against Pulsar effortlessly and without affecting performance. There is nothing like it on the pub-sub ecosystem.
In this short session, we will revisit what Pulsar SQL is, how to make the best out of it, how to deploy it, and how to use it!
Migrate to Serverless with Pulsar Functions
How many times have you needed to do some transformations between topics but deploying a whole service means problems? Or you thought of using AWS Lambda functions but they are extremely expensive?
Since Pulsar 2.0 we have Pulsar Functions to have a native, serverless, lightweight computing option in order to fill such a need in a couple of hours instead of a couple of weeks of work!
In this session, we will explore what Pulsar Functions are, why they can leverage your needs, and how we can develop them easily to migrate full applications and ETLs into serverless within Apache Pulsar.
Axel Sirota
Founder @ Data Trainers
Buenos Aires, Argentina
Links
Actions
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top