Many websites have a way to search through listings. However the UI often includes many filters, check boxes, dropdowns and other UX nightmares. What if the user could just write in plain English what they wanted to find and get suitable results? We explore it as we build a version of linkedin's AI job search in Django using LangChain!

Python is at the core of our analytics platform, which processes over 8,000 game records daily, each approximately 500 MB in size. Over the past two years, we have accumulated more than 200 TB of data, equivalent to 1,600 years of game time from over 7 million players—and our goal is to increase this user count tenfold. This talk will cover how we transitioned from Go and C++ parsers connected via PyBind to data frames in Python, how our analyses evolved from Pandas to Polars, and why we migrated our backend from Django to FastAPI. Finally, we will share our real-world experience with performance optimization, leveraging RabbitMQ, Redis, and process monitoring in an environment where Python bridges the worlds of game data and AI analysis.