Skip to content

Releases: DivakarUngatla/wayfinder

Wayfinder v0.4.0 — LLM-as-a-Judge Evaluation

Choose a tag to compare

@DivakarUngatla DivakarUngatla released this 21 Aug 20:02

🚀 Wayfinder v0.4.0 — LLM-as-a-Judge Evaluation

This release accompanies AI Engineering Fundamentals – Part 5: LLM-as-a-Judge.

What's Included

  • LLM-as-a-Judge evaluation framework
  • Configurable evaluation criteria and scoring rubrics
  • LLM-based response evaluation with structured outputs
  • Judge prompts for evaluating AI responses
  • Representative LLM evaluation dataset
  • Local evaluation runner
  • LangSmith evaluation integration
  • Updated project documentation and README

This release represents the code used throughout the article.

Wayfinder v0.3.0 — Human Evaluation

Choose a tag to compare

@DivakarUngatla DivakarUngatla released this 08 Aug 16:42

🚀 Wayfinder v0.3.0 — Human Evaluation

This release accompanies AI Engineering Fundamentals – Part 4: Human Evaluation.

What's Included

  • Human evaluation framework
  • Representative evaluation dataset
  • OpenAI-powered response generation
  • Human evaluation workflow
  • Evaluation rubrics and scoring guidelines
  • Markdown-based evaluation reports
  • LangSmith integration
  • Updated project documentation and README

This release represents the code used throughout the article.

Wayfinder v0.2.2 — Rule-Based Evaluation

Choose a tag to compare

@DivakarUngatla DivakarUngatla released this 21 Jul 18:07

🚀 Wayfinder v0.2.2 — Rule-Based Evaluation

This release accompanies AI Engineering Fundamentals – Part 3: Rule-Based Evaluation.

What's Included

  • Rule-based evaluation framework
  • Ground truth evaluation dataset
  • Local evaluation pipeline
  • LangSmith integration
  • Improved evaluation examples and console output
  • Architecture diagram
  • Updated project documentation and README

This release represents the code used throughout the article.

Milestone 2 – Rule-Based Evaluation

Choose a tag to compare

@DivakarUngatla DivakarUngatla released this 17 Jul 20:19

🚀 Milestone 2 – Rule-Based Evaluation

This milestone introduces the first evaluation framework for Wayfinder.

Highlights

  • Rule-based evaluator
  • Local evaluation example
  • LangSmith integration
  • Ground truth evaluation dataset
  • Overall pass/fail scoring

This milestone demonstrates how deterministic evaluation can validate AI application behavior before introducing more advanced evaluation techniques.

Milestone 1 – Basic Flight Search

Choose a tag to compare

@DivakarUngatla DivakarUngatla released this 17 Jul 20:16

🚀 Milestone 1 – Basic Flight Search

The first milestone of Wayfinder introduces the core application architecture.

Highlights

  • Flight domain models
  • Flight search service
  • Search flight tool
  • Wayfinder agent
  • Runnable example

This milestone establishes the foundation that future AI engineering concepts will build upon.