Work & Methodology

My focus is on automating workflows to build reliable systems. Here's how I approach AI development.

Evaluation-Driven Development

Applying testing principles to AI.

My background in testing taught me the importance of determinism. When I started working with AI, I realized we needed similar rigor. I treat prompt engineering like software development: every change should be backed by an evaluation. Using tools like LLM-as-a-judge and frameworks like DSPy helps ensure our agents are reliable and consistent.

Back prompt changes with evaluation metrics
Use LLM-as-a-judge for nuanced quality assessment
Optimize prompts systematically with frameworks like DSPy
Treat agent behavior like test cases—aim for reproducibility
Verify changes with evaluations before shipping

Addressing Limitations in RAG

Standard RAG approaches can struggle with time-sensitive information or connecting dots across many documents. I've been working on a memory framework that tries to address these gaps, ensuring agents have the right context when they need it.

Cursor Community Leadership

I've been fortunate to be an active member of the Cursor community, where I help others explore AI-assisted coding. I host regular workshops to share what I've learned about building agents and optimizing workflows.

Cursor Ambassador2000+ Hours AI CodingThousands Trained

Impact Story

Mike Miquela

Exploring New Tools

Before

Mike, a colleague and experienced engineer, was initially skeptical about how much AI tools like Cursor could actually help his workflow.

After

We spent some time exploring the features together, and he found it really valuable. He's since deep-dived into the tech, started sharing his own learnings at conferences, and is now doing great work building similar tools at Salesforce.

"It was amazing to see him embrace the tool and run with it."

Open Source

Professional Experience

Senior AI Software Engineer

May 2023 – Present
Vivun
  • Developing multi-agent systems using LangGraph, exploring voice and avatar interfaces.
  • Building deep research agents and workflow orchestrators that leverage shared long-term memory.
  • Working on memory architectures using knowledge graphs to improve context retrieval beyond standard RAG.
  • Implementing evaluation-driven development to ensure prompt changes are verified systematically.
  • Creating workflows in Cursor to help the team move faster.
LangGraphPythonFastAPINextJSLLMsRAGKnowledge GraphsDSPy

Cursor Ambassador

Mar 2025 - Present
Cursor
  • Connecting with the community to share feedback with the Cursor team.
  • Hosted workshops to help other developers get the most out of the tool.
  • Sharing my own experiences and workflows for building full-stack apps.
  • Helping peers discover and adopt new AI workflows.
CursorAI ToolsCommunityDevRel

Lead SDET / Test Architect

June 2021 – May 2023
Priceline
  • Designed test automation frameworks to improve reliability and scalability.
  • Worked on CI/CD pipelines to streamline the release process.
  • Contributed to backend services using Spring Boot.
  • Learned the value of thorough testing, which influences my current work.
JavaScriptNode.jsWebDriverIOGitHub ActionsSpring Boot

Automation Developer

Sep 2020 – June 2021
Powerhub Inc.
  • Built test automation using Cypress.
  • Helped improve code quality with linting tools.
  • Conducted performance testing for APIs and UIs.
CypressMochaJavaScriptMongoDB

Quality Assurance Automation Engineer

Feb 2019 – Apr 2020
NXM Labs INC
  • Tested systems across Blockchain, Mobile, and IoT.
  • Built API tests for IoT devices.
SeleniumPythonJavaIoTBlockchain

Software Tester

May 2014 – Apr 2016
Webyoutechnologies
  • Started my journey in QA, building Selenium test cases.
  • Learned the fundamentals of software quality.
SeleniumTestNGJava