Tame Messy Data, Build Smarter Agents

Data Meets Intelligence, IndyPy’s November 2025 meetup, focused on two things Python teams need right now: how to standardize messy data and how to make agentic AI behave predictably in real systems.

The theme across both talks was consistent: reliability doesn’t happen by accident. It comes from strong boundaries (validated data) and strong guardrails (observable, well-scoped agent workflows).

Pydantic Brings Order to Data Chaos

Kyle Adams (Staff Software Consultant at Test Double) opened with a familiar problem. Inputs arrive incomplete, inconsistent, and occasionally cursed. He walked through how Pydantic provides a structure-first approach, so services stay stable even when the outside world isn’t.

Highlights:

• Validate and serialize data safely across service boundaries
• Normalize inconsistent payloads with aliases, nested paths, and generators
• Model domain objects that scale across APIs and microservices
• Avoid common pitfalls like by_alias confusion and config traps
• Put Pydantic to work in modern stacks, including database and AI pipelines

Watch the full talk:

AI Agents Without the Hype Cycle

Robert Herbig (Lead Software Engineer & AI Practice Lead at SEP) tackled what actually makes something an AI agent, and why so many fall short. He broke down agents as loop-driven workflows (plan → act → observe → adjust), then focused on the real work: integrating tools and internal data without turning reliability and security into afterthoughts.

Highlights:

• Know when an agent is the right tool (and when it isn’t)
• Implement core agent patterns with Python and LangGraph
• Connect tools safely with MCP and add human-in-the-loop guardrails
• Instrument agents with observability (LangSmith, Jaeger, Langfuse)
• Evaluate agent behavior so it stays predictable and secure

Watch the full talk:

With the right tools, teams can tame messy data and build AI systems that behave predictably, helping them ship reliable features faster and reduce surprises in production.

Curious what’s coming next? View upcoming IndyPy events: https://www.meetup.com/indypy/.