Coffee & Code Philly

# Open-Source AI in Practice: Build and Evaluate Agentic Tools with OpenLLM

Join Coffee & Code Philadelphia and OpenLLM for a hands-on workshop focused on building, evaluating, and contributing to open-source AI tools.
This session is designed for developers who want practical experience working with open-source models, agentic workflows, and the infrastructure required to operate AI systems reliably.
Rather than spending the day on high-level presentations, we will focus on one technical theme, a live demonstration, and a guided exercise that participants can follow or adapt to their own projects.

## What We’ll Cover

During the workshop, participants can expect:

• A practical introduction to OpenLLM and the problems it helps developers solve
• A live demonstration of an open-source AI model, tool, or agent workflow
• A guided exercise to build or modify a small AI application
• Techniques for evaluating agent behavior, reliability, and output quality
• Discussion of safety, deployment, and operating open-source AI in real projects
• Examples of how developers can contribute code, documentation, integrations, testing, and feedback
• Open Q&A with the OpenLLM team and Philadelphia builders

The exact technical focus and exercise will be announced closer to the event.

## Who Should Attend

This workshop is intended for:

• Software engineers
• AI and machine learning builders
• Technical founders
• Researchers and students
• Open-source contributors
• Developers experimenting with AI agents, models, evaluation, or infrastructure

Participants of different experience levels are welcome, but basic programming experience will be helpful for the hands-on portion.

## What to Bring

Please bring:

• A laptop
• A development environment you are comfortable using
• A GitHub account
• Any installation instructions or prerequisites announced before the event

You are also welcome to bring an existing AI or open-source project you are working on.

## Why This Event

The goal is to give Philadelphia builders direct experience with OpenLLM and its broader open-source ecosystem while creating space for useful technical feedback and collaboration.
This workshop may also serve as the starting point for future OpenLLM and Coffee & Code Philadelphia programming, including recurring workshops, contributor sessions, hackathons, and other open-source AI events.

## About OpenLLM

OpenLLM is an open-source initiative focused on making it easier for developers to build, run, evaluate, and share applications powered by open models.
Learn more at openllm.sh.

## About Coffee & Code Philadelphia

Coffee & Code Philadelphia is a community of approximately 5,000 developers, founders, students, researchers, and technology professionals.

## Claude Code vs. Codex: Loops, Graphs, and Harnesses

This is a casual pair-programming meetup for developers experimenting with Claude Code, Codex, and agentic coding workflows.
There is no formal presentation and no fixed curriculum. Bring a laptop, a repository, a side project, or a workflow you want to test. We will work in small groups, compare approaches, debug problems, and share what is working.

Possible areas to explore include:

• Building iterative coding-agent loops
• Comparing Claude Code and Codex on the same task
• Designing graph-based agent workflows
• Creating reusable coding harnesses
• Managing context across large repositories
• Running tests and validating agent-generated changes
• Using sandboxes, permissions, and approval steps
• Coordinating subagents or parallel tasks
• Measuring latency, token usage, cost, and completion quality
• Debugging failed or incomplete agent runs

You do not need to be an expert in either tool. This meetup is for people who want hands-on time to experiment, pair with other developers, and learn through building.
Bring something you are working on, or join someone else’s project.
Hosted by Coffee & Code Philadelphia.