OpsPilot

AI-Assisted Incident Operations

An AI-assisted IT operations platform designed to help technical teams analyze incidents, generate structured recommendations, and safely act on them.

OpsPilot integrates with ServiceNow through REST APIs and combines AI analysis with validation, explicit human approval, controlled write-back, read-only safeguards, and PostgreSQL audit logging.

TypeScript · AI · ServiceNow REST API · PostgreSQL · Neon · ITSM

Meal Mastery

From Side Project to Published Product

A mobile meal-planning application built with React Native and TypeScript, designed around making weekly meal planning, pantry management, recipes and shopping easier.

Built and shipped independently from concept through production release.

React Native · Expo · TypeScript · AsyncStorage · Android

SuperR — AI-Assisted IT Troubleshooting

An AI troubleshooting agent designed to help Tier-1 support technicians solve problems faster, take fewer unnecessary actions, and escalate with better evidence.

SuperR was built for the micro1 2026 Hackathon as an experiment in applying AI to a familiar enterprise IT problem: troubleshooting efficiently without sacrificing human judgment.

Rather than simply generating troubleshooting suggestions, SuperR evaluates the available evidence, selects targeted diagnostic actions, determines whether an issue can be safely resolved or should be escalated, and produces a structured support ticket containing the symptoms, evidence, disposition, and appropriate assignment group.

The Problem

Traditional troubleshooting often relies on broad checklists. Technicians may perform numerous diagnostic and remediation steps before identifying the actual problem.

SuperR explores a different approach:

Can an AI agent reach the correct support decision while performing substantially fewer actions?

Measured Results

On a five-case holdout evaluation:

• 100% disposition accuracy
• 83.3% reduction in troubleshooting actions vs. the deterministic baseline
• 3.0 average actions per case vs. 18.0 for the baseline
• 0 unnecessary remediation actions
• 100% ticket completeness

On the 12-case development evaluation, SuperR Agent v2 reduced total troubleshooting actions by 78.2% while maintaining 100% accuracy.

How It Works

SuperR combines:

Python • OpenAI API • Streamlit • Structured JSON Cases • Deterministic Evaluation

Each simulated support incident contains observable system state and diagnostic evidence. The agent chooses which evidence to investigate instead of receiving every answer upfront.

Its final output includes:

• Resolve or escalate decision
• Root-cause classification
• Targeted remediation when appropriate
• Assignment group for escalations
• Structured ticket title
• Symptoms and supporting evidence

A deterministic baseline, development cases, and a separate holdout set make it possible to measure whether the AI agent actually improves the troubleshooting process rather than simply appearing intelligent.

Why I Built It

After years of working in enterprise IT support, I’ve seen that the hardest part of troubleshooting often isn’t knowing more commands—it’s knowing which action is worth taking next.

SuperR explores how AI can augment that judgment while keeping the technician in control.

The goal isn’t to replace Tier-1 support.

It’s to give Tier-1 technicians better tools.

Tech Stack

Python • OpenAI API • Streamlit • JSON • Agentic Troubleshooting • Automated Evaluation • Structured Ticket Generation