Daniel Dash

Senior Quality Development Engineer, InterSystems

Building AI-assisted engineering systems, test infrastructure & automation for InterSystems IRIS

I build infrastructure-backed automation and AI-assisted systems for complex engineering workflows: regression automation, failure investigation, codebase analysis, and practical AI adoption.

Role Thesis

A senior engineering role at the edge of workflows and platform capabilities

My work sits between users and platform capabilities. At InterSystems, that means translating Quality Development testing, regression, and environment needs into automation workflows built on internal VM provisioning, InterSystems IRIS deployment, and environment-configuration infrastructure. That infrastructure is not AI infrastructure. Alongside the formal automation role, I have proactively driven AI-assisted engineering work across QD, including failure investigation, codebase impact analysis, reusable AI/MCP tooling, mentoring, and adoption enablement.

Featured Work

Formal automation ownership and self-initiated AI engineering work

Evolve and maintain production systemRegression harnessIRISInfrastructure automationPlatform adoption

Regression Harness Engineering for InterSystems IRIS

Owning a 10,000+ test regression harness that translates QD regression testing needs into TAE-backed automation workflows.

This is Daniel's formal infrastructure-backed automation work. He owns a large-scale InterSystems IRIS regression testing harness used to run 10,000+ tests across active branches. The work translates Quality Development requirements into automation built on TAE capabilities for VM provisioning, IRIS deployment, environment configuration, and job execution. On top of that infrastructure layer, Daniel built custom resource querying and management, job queueing, platform targeting, failure lifecycle automation, and test exclusion/inclusion governance workflows. The harness reduced job errors by approximately 80% and reduced manual triage from 70%+ of active time to under 10%.

Shows

End-to-end ownership, platform fluency, cross-functional coordination, reliability instincts, and translation from ambiguous engineering needs to operational systems.

Maturity: Evolve and maintain production system

maintainedLLM workflowRAGStructured outputsSME review

Evidence-Grounded AI Failure Investigation for InterSystems IRIS Regression Workflows

A deployed and maintained AI workflow that turns regression failure artifacts into evidence-grounded investigation reports for SME review.

This self-initiated AI-assisted workflow addresses a formerly manual investigation process for IRIS regression failures. It assembles preprocessed unit test logs, preserved application and system logs, deterministic test code context, version-aware documentation retrieval, historical test result signals, and potentially relevant product changes. The system produces structured investigation reports with failure type classification, evidence citations, log correlations, failure signatures, and remediation steps, then enriches Jira tickets for subject-matter expert review. It reduces active SME analysis from minutes-to-hours of synchronous work to a few asynchronous minutes while keeping remediation decisions human-owned.

Shows

Practical AI deployment judgment: context engineering, retrieval quality, evals, guardrails, ticket integration, and reviewer control.

Maturity: Deployed and maintained AI-assisted workflow

builtObjectScriptCompiler signalsMCPDependency graph

Codebase Impact Graph for Safer Changes in ObjectScript Systems

A locally used ObjectScript dependency graph tool that gives engineers and AI coding agents deterministic codebase context.

As AI coding agents become more common, ObjectScript-heavy systems need better dependency context than generic code graph tools, grep, or oversized prompts can provide. This internal tool builds a deterministic call-dependency graph using source parsing plus IRIS compilation, so macro expansion, generated methods, inherited methods, and compiler-visible relationships can be surfaced at the symbol level. Engineers can query it from the command line or expose it through MCP for coding agents. It is locally used by Daniel and some other engineers, not a hosted service or automated production pipeline.

Shows

Ability to build deterministic context systems that make AI coding workflows safer, more inspectable, and more token-efficient.

Maturity: Built and locally used internal tool

What I Build

Practical systems for complex engineering workflows

System 01->

Deployment engineering for test infrastructure adoption

System 02->

AI-assisted engineering workflows

System 03->

Platform targeting and environment automation

System 04->

Codebase analysis and impact reasoning

System 05->

Practical AI adoption and enablement

System 06->

Evals, guardrails, and human review patterns

Evidence

Selected public-safe scale signals

10,000+
Regression scale

Tests per large-scale regression run.

5-30
Typical triage surface

Failures per run requiring recurrence analysis, exclusion governance, or bug association.

~80%
Harness reliability

Reduction in regression harness job error rate through managed resource querying and queueing.

<10%
Manual triage load

Reduced from 70%+ of active time through failure lifecycle automation.

Operating Pattern

How I tend to make ambiguous work reliable

Across my work, the pattern has been translating ambiguous engineering needs into reliable automation: understanding the workflow, mapping requirements to platform capabilities, making pragmatic design tradeoffs, and owning systems through implementation, adoption, and iteration.

Writing

Short notes on practical AI workflow reliability

Contact

Open to AI deployment, AI engineering, and field-facing engineering roles

I’m especially interested in work that combines user discovery, platform fluency, implementation ownership, eval discipline, and pragmatic AI deployment.

Contact