About
Software engineer.
I've built and shipped backend services, APIs, and frontend features across production environments.
More recently, I've been focused on using AI to improve developer workflows and automate parts of the development process.
I'm interested in building tools that make complex or messy behavior easier to understand and work with.
Experience
Where I've worked
Software Developer
Fusion92
- →Built an AI-powered E2E test failure analyzer for a 140+ test Playwright suite — combining GPT-4o vision for screenshot analysis with a LangChain tool-calling agent to automate root cause investigation.
- →Reduced manual debugging time from ~2 hours to under 5 minutes per run via structured, categorized failure attribution reports (test vs. application fault).
- →Shipped a multi-tenant Chatbase AI chatbot integration within CoreCMS with tenant-safe isolation, secure secret storage, and HMAC-based session validation.
- →Implemented an admin provisioning utility and agent-guided onboarding tour (Shepherd.js) with CMS-managed tooltips and restart/skip controls.
Full Stack Developer Intern
Fusion92
- →Led redesign of the Platform Management page in CoreCMS using Vue 3 components, improving usability and UI structure.
- →Collaborated cross-functionally with product and client teams in an Agile/Jira environment to scope and ship features.
- →Validated APIs via Swagger and supported QA and release processes through Azure DevOps pipelines.
Recent Project
FailChain
End-to-end test failure analysis tool. Parses a test run, groups failures by error signature, and generates a structured report with root cause, fault classification, and recommendations.
Before
Playwright E2E run, test failure with no root cause context
Failure attribution
Debug output
Agent reasoning
Flaky test detection
Architecture
Key details
- →Framework-agnostic — designed to work across E2E testing frameworks without coupling to any one tool
- →Vision model analysis per failure group + LangChain tool-calling agent for root cause investigation
- →Static pre-analysis pass runs before the LLM — catches obvious issues deterministically and cheaply
- →Classifies failures as test fault, application regression, or flaky — per run
- →Failure history tracked across runs to surface recurring flakiness patterns
- →Full step-by-step audit log for every reasoning chain
Skills
Technical foundation
Languages
Frameworks
AI / Systems
Infrastructure
A bit about me
Outside the code
Outside of work I'm usually hooping, out on a trail, or deep in a game. Always building something on the side too.
