Skip to content

Program

AI & Agents Channel

Curate the practical use of AI engineering — agents, RAG, evals, fine-tuning — across the community.

Quarterly deep-dive meetups, ongoing async threads.

Contact usHow to join

The AI & Agents Channel is the place where Pereira's AI builders share what is actually working in production. Async conversation in the community channel, plus quarterly deep-dive meetups. Bias toward demos and code, away from hype.

Why it exists

In 2024 the community’s conversation shifted. Every meetup ended with questions about LLMs, agents, RAG, evals. Some people were experimenting, others were hesitating, almost nobody was sharing what they already had running in production.

The AI & Agents Channel was created to change that dynamic: a dedicated space where people share what works, what broke, and what turned out to be smoke. The unwritten rule: if you are going to tell us about it, bring the demo or the code.

The bet: the people in Pereira shipping AI to production are mostly invisible to each other. Make the channel exist and they will find each other.

What the channel covers

  • LLMs in production — real costs, real latencies, real evals.
  • Agents and agentic patterns — tool use, planning, multi-step, MCP, A2A.
  • RAG and memory — chunking, embeddings, ranking, relevance evaluation.
  • Fine-tuning and adaptation — when it is worth it, when it is not.
  • Eval and observability — how to know whether your system actually improved.
  • Prompts vs. engineering — the difference between a clever prompt and a robust system.
  • Applied ethics and security — bias, leaks, PII handling, prompt injection attacks.
  • The stack we use — Claude, GPT, Gemini, open-source models. Frameworks and tools, with opinions.

Format

Two parallel rhythms:

Ongoing async conversation

  • In the community channels (mainly LinkedIn, plus an internal group conversation).
  • Anyone can ask a question, share a demo, or open a discussion.
  • The rule: bring data or code. Pure-opinion threads get redirected, kindly.

Quarterly deep-dive meetups

  • Once a quarter, a full meetup dedicated to the channel.
  • Format: 2-3 technical talks (45 min each) + a panel with questions + live demos.
  • Recent topics: agents for product management, procedural generation and drug discovery, generative AI applied to internal workflows.

How to join

  • If you work with AI in production (or close to it): write to us through the contact form with the subject “AI Channel” and tell us what you are building. We will connect you.
  • If you want to learn: subscribe to the meetups via Luma and filter by the AI / Agents tag.
  • If you want to present: the bar is technical honesty, not seniority. If you have built or tested something — even if it failed — we want to hear it.

What we are NOT

  • We are not a hype group. We do not argue about which model won the latest synthetic benchmark.
  • We are not vendor evangelism. We show what we use, with real costs, latencies and trade-offs.
  • We are not a course. There is no introductory material. The general monthly meetups and the Speaker School already cover that.

Published resources

The channel’s talks and demos are published as part of the site catalog (Spanish and English):

FAQs

Do I have to be senior? — No. The bar is bringing something concrete: a demo, some code, a specific question.

Is it only about Claude/GPT? — No. We cover the whole relevant ecosystem, including open-source models.

Are the sessions recorded? — Yes, whenever possible, and they are published in the slides catalog and the blog.

What language is it in? — Mainly Spanish. Slides and notes are published in both Spanish and English.

Related meetups