Automation & AI Engineering

Building systems that actually work for real families

I design and build AI-powered automation systems that solve genuine pain points—from voice assistants to family logistics platforms. Every project starts with a real problem and ends with something people actually use.

6+Production Systems
12+API Integrations
5Daily Active Users

Selected Projects

Voice AI

Jarvis

Distributed voice assistant with family-aware context switching

The Problem

Commercial voice assistants are generic and disconnected from family context. They can't remember that Riley has a math test tomorrow or that Alex prefers Boston sports updates.

The Approach

Built a local-first voice assistant that runs on distributed hardware (N100 mini PC → Raspberry Pi nodes), with per-family-member profiles and context awareness.

Key Decisions
  • Local-first architecture for privacy and speed
  • Wake word detection via Picovoice
  • ElevenLabs for natural-sounding TTS with voice cloning
  • Deepgram for fast, accurate STT
  • Family profiles stored locally, context injected per-request
Outcomes

Working prototype with touchscreen UI, multiple room support planned, and integration points for calendar, weather, and school schedules. Daily use for morning briefings.

Tech Stack
PythonClaude APIElevenLabsDeepgramPicovoiceRaspberry PiFlask
SMS-First Product

GiftStash

Capture gift ideas via text message, organized automatically

The Problem

Great gift ideas come at random moments—overhearing a comment, a text from someone. By the time you need them, they're forgotten. Apps require too much friction.

The Approach

SMS-first design: text a gift idea to a number and it's automatically parsed, categorized by recipient, and stored. No app to open, no login required in the moment.

Key Decisions
  • A2P 10DLC compliance for reliable SMS delivery
  • Claude for natural language parsing (handles "mom would love those blue earrings from Nordstrom")
  • PWA for web access when browsing gift lists
  • Chrome extension for one-click capture while browsing
Outcomes

Family actively using for birthday and holiday planning. Captured 50+ gift ideas that would have been lost. Zero-friction input has transformed how we track gift inspiration.

AI-Generated Assets

Custom avatars and illustrations created with Gemini for user profiles and marketing.

Avatar Avatar Avatar Avatar Illustration
Tech Stack
TwilioClaude APIVercelReactPWAChrome ExtensionGemini (Images)
Data Integration

School District Dashboard

Unified view of multiple schools' calendars, events, and communications

The Problem

With three kids across multiple schools, information comes from dozens of sources: emails, apps, websites, paper handouts. Critical dates get missed.

The Approach

Aggregate and normalize data from multiple school sources into a single dashboard with smart filtering by child, event type, and urgency.

Key Decisions
  • Email parsing for school newsletter extraction
  • Calendar aggregation across school platforms
  • Child-specific views for quick daily reference
  • Integration with family DAKboard display
Outcomes

Single source of truth for school logistics. Reduced "I didn't know about that" moments. Morning briefings now include personalized school updates.

Tech Stack
PythonGmail APICalendar APIsReactDAKboard
AI Automation

Email Priority Inbox

AI-powered email triage that surfaces what actually matters

The Problem

Email volume makes it impossible to distinguish urgent items from noise. Important messages from schools, family, or time-sensitive requests get buried.

The Approach

AI-powered classification that understands personal context: who matters, what topics are urgent, and what can wait or be automated away entirely.

Key Decisions
  • Context-aware priority scoring (not just sender rules)
  • Integration with family calendar for deadline awareness
  • Digest generation for low-priority batches
  • Learning from manual overrides
Outcomes

Reduced email processing time significantly. Critical school communications now surfaced immediately. Marketing and low-priority items auto-sorted.

Tech Stack
Claude APIGmail APIPythonVercel
AI Education

Study Buddy

AI-powered flashcard app designed for how kids actually learn

The Problem

Traditional flashcard apps are boring and don't adapt to individual learning patterns. Kids lose interest quickly, and parents can't easily create good study materials.

The Approach

AI generates age-appropriate flashcards from any topic, adapts difficulty based on performance, and keeps engagement high with encouraging feedback.

Key Decisions
  • Prompt engineering for kid-friendly explanations
  • Spaced repetition algorithm for retention
  • Parent dashboard for progress tracking
  • Voice input option for younger kids
Outcomes

Used weekly for spelling, math facts, and science vocabulary. Kids actually ask to use it—the AI encouragement and adaptive difficulty keep them engaged.

AI-Generated Mascot: CottleCub

Created a friendly bear mascot with 25+ expressions for different subjects, moods, and break activities.

CottleCub CottleCub CottleCub CottleCub CottleCub CottleCub
Tech Stack
Claude APIReactVercelWeb Speech APIGemini (Images)
Operations System

Family HQ

Notion-based command center for household operations

The Problem

Running a household with three kids involves hundreds of recurring tasks, decisions, and information that lives in different people's heads or scattered systems.

The Approach

Centralized Notion workspace that serves as the single source of truth for schedules, contacts, recurring tasks, kid info, vendor details, and household SOPs.

Key Decisions
  • Relational databases linking kids → activities → schedules → contacts
  • Template system for recurring events (birthday party planning, etc.)
  • Mobile-friendly views for quick access
  • Integration points for automation systems
Outcomes

Eliminated "where's that info?" moments. New babysitters can find emergency contacts in seconds. Household runs more smoothly with less mental overhead.

Tech Stack
NotionNotion APIZapierApple Shortcuts
About

Building with intention

I approach automation and AI the same way I approach any engineering problem: start with a real pain point, understand the constraints, and build something that works reliably in the real world.

My background in measurement and analytics taught me to focus on outcomes, not just outputs. Every system I build has clear success criteria and gets iterated based on actual usage.

Currently exploring what's possible when you combine modern AI capabilities with thoughtful systems design for personal and family use cases.

  • AI Integration

    Claude API, prompt engineering, context management, multi-model architectures

  • Automation Systems

    Apple Shortcuts, Zapier, custom integrations, event-driven workflows

  • Full-Stack Development

    React, Python, Vercel, API design, database architecture

  • Voice & Conversational AI

    ElevenLabs, Deepgram, Picovoice, conversational design

  • AI Image Generation

    Gemini, prompt crafting for consistent character design and brand assets