Pratik Patel
Lead Developer Relations Engineer
Atlanta, Georgia, United States
Actions
Pratik Patel is a Java Champion, developer advocate at Azul Systems and has written 3 books on programming (Java, Cloud and OSS). An all around software and hardware nerd with experience in the healthcare, telecom, financial services, and startup sectors. He's also a co-organizer of the Atlanta Java User Group and Enterprise AI Atlanta, conference chairperson for Devnexus, frequent speaker at tech events, and master builder of nachos.
Area of Expertise
Topics
Your GPU, Your Model: Running Open-Weight AI
Tired of cloud API latency, rate limits, and per-token bills? Open-weight models have gotten good enough, and runtimes fast enough, that your workstation can do real work. This session covers how to run, quantize, and serve open-weight models on your own hardware, and where the limits are.
We'll start with what a model actually is (weights, parameters, tokens) and why running it locally makes sense (or doesn't). Then we'll get practical:
Quantization: choosing a method and bit width for your memory budget, and what you give up
Inference engines: picking and configuring the right runtime for your hardware and use case
Hardware sizing: rules of thumb for memory bandwidth, VRAM, and context window management on local silicon
You'll leave knowing what your machine can realistically run, how fast, and when the cloud is still the better call. Bring your questions.
Fine-Tuning an LLM to Cut Costs and Go Faster
Your company has millions of product reviews and a product team asking one question: what are customers actually complaining about? Star ratings won’t tell you. Sending every review to a frontier LLM will tell you, but at a cost and latency nobody will approve for a weekly report.
In this hands-on workshop, you’ll build the pattern teams use in production: a big model labels, a small model scales, and SQL answers the question. Starting from a public dataset of Amazon product reviews on the Hugging Face Hub, you’ll use DuckDB to sample and explore the data. Then you’ll have a large LLM tag a few thousand reviews with product aspects (battery, fit, durability, shipping) and sentiment. You’ll LoRA fine-tune a small open LLM to return those labels as JSON, run it over the full dataset, and write the results back to Parquet on the Hub.
Then the fun part: with every review turned into structured columns, you’ll use DuckDB to answer real Voice-of-the-Customer questions. Which brands get the most “arrived broken” complaints? Did battery complaints spike after a product revision? Finally, you’ll ship the pipeline as an interactive app on Hugging Face Spaces.
Along the way, you’ll learn when fine-tuning is the right tool and when it isn’t. Fine-tuning teaches a model a task, not facts. That’s why it’s the right tool for turning messy text into structured data, and the wrong one for helping a model “know” your reviews.
You’ll leave with a working, reusable pipeline and a clear mental model for choosing between prompting, RAG and fine-tuning.
Data Isn't the New Oil — It's Refined Data
“Data is the new oil,” but crude oil is worthless until it’s refined. AI now sits inside the software delivery pipeline: coding assistants, internal chatbots, and agents that write, review, and ship code. They’re only as good as the data behind them, and most company data is crude: duplicated, stale, and noisy. This talk explains what refining data really means and how it raises the quality of the software AI helps deliver. You’ll learn to treat data like code (versioned, tested, and gated in CI) so that better data turns into better software.
AI Native Architecture for Java Applications
We are currently moving from "AI-enabled" systems, where artificial intelligence is an additive feature, to "AI-native" systems, where intelligence is the foundational, architectural core. An AI-native application is not merely a traditional application with a machine learning model bolted on; it is an entirely new class of software designed from the ground up to learn, adapt, and act autonomously. These systems are architected around continuous data ingestion, real-time model interaction, and a contextual understanding of the application and runtime environment.
We'll discuss the key elements mentioned above and the difference between existing applications that have added AI capability as an accessory, and this new class of applications that are built with AI in mind from the start. While we'll use the Java ecosystem in examples, the principles we'll discuss are language agnostic. We'll focus on the architecture discuss these topics:
* Big Data and data pipelines
* Integration with other services and APIs
* Testing considerations
* Evolution of UI/AX
Big Data and AI Architecture: Apache Iceberg, Spark and LLMs
This presentation delves into the potential of integrating LLMs with Apache Spark and Apache Iceberg as part of a Big Data to AI foundational architecture. In this session we’ll explore the potential of combining Iceberg, Spark and LLMs to give you a real world AI architecture that uses your data.
We'll build an AI application that allows users to perform data queries and extract insights from massive datasets using natural language. We'll start with understanding the structure and architecture of a large dataset. Then we'll look at options for querying the dataset using Apache Spark and Trino. Finally, we'll use an LLM to query the dataset using natural language. We'll also look at other uses of LLMs as part of an overall solution, and explore the differences between different LLMs.
We’ll also discuss where event streaming (Kafka and Flink) fit into this architecture. The design of this architecture is meant to be flexible and give your dev team the ability to choose different technologies for the processing and querying. I’ll leave you with a CONCRETE example that you can run on your laptop and explore the possibilities. Again, this will be an example of a real-world application; the dataset used will be for home sales data for the last 15 years.
We will use these technologies:
* Apache Iceberg
* Apache Spark
* Spring AI or LangChain TypeScript/Python
* LM Studio
* Various LLMs
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