Papaya to Host Tel Aviv Meetup on AI Production Stacks and Cost Control
Papaya is hosting a Tel Aviv meetup on November 9 to address AI integration challenges, LLM cost management, and the balance between language models and traditional code.
A professional meetup titled "Own Your AI Stack" is scheduled for November 9 at Papaya's offices in Tel Aviv, aiming to address the engineering challenges, spiraling costs, and performance bottlenecks of integrating Large Language Models (LLMs) into production environments. Jointly organized by Papaya and local tech media, the event targets developers, engineering managers, architects, and data professionals.
The meetup will explore the critical boundary between deploying LLMs and returning to traditional code, presenting architectural solutions for internal AI tools. Organizers noted that while companies increasingly adopt tools like Claude and Cursor, they often encounter issues with high costs, slow response times, and difficulties processing large volumes of data.
Practical Insights and Case Studies
The agenda features two main technical sessions addressing real-world implementation hurdles. The first presentation will examine why not every task requires an LLM, detailing behind-the-scenes challenges encountered while implementing autonomous agents for Meta campaign management and creative decoding. Speakers will discuss architectural shifts required to overcome performance obstacles.
"The second lecture will showcase Clodi, an in-house desktop alternative to Claude developed at Papaya," organizers stated. Originally created as a solo experiment after transitioning from Cursor to Claude via AWS Bedrock to cut costs, the terminal-based tool proved insufficient for developers needing graphical interfaces, such as Unity teams. This led to the creation of a custom desktop client running on the Claude Agent SDK without a subscription, which has since become an essential daily tool for the company's developers.
Key Takeaways for Engineers
Attendees will gain actionable knowledge on optimizing AI integration, including:
-
Decision Frameworks: Methodologies for determining when to deploy LLMs versus traditional APIs to conserve resources.
-
Observability: Integrating Langfuse from inception for data-driven debugging and transparency.
-
Cost Management: Strategies for handling session caching, long-context token budgets, and migrating to AWS Bedrock.
-
Internal Tooling: Blueprints for building custom wrappers, team plugin systems, and deterministic media processing.
Participation is free of charge, but advance registration is required due to limited seating.
