AI PORTFOLIO 2026

Nico Toneff

Senior Technical Program Manager. A curated selection of tools I've built with AI, at work and on my own time.

SAVE IT AS PDF
01

Jira Flow Timelapse

See what actually happened in Jira as an animated 'instant-replay' over any time period.

HOW I USED AI

I specified the model: an Epic kanban with a task-level sub-kanban, replayed from Jira's own history. Cards fly column to column and reshuffle by priority; completed tasks light their Epic and chime, so you hear throughput and lulls as well as see them. Claude wrote the changelog parser and the animation renderer.

The board mid-replay, cards moving between columns as the history plays
The configure step: mappings pre-filled from Jira's status categories, then dragged
BOARD ISSUE TYPE

Which issue type becomes a board card

CUSTOM CARD COLORS

Age, stagnation, active time, or any field

CARD SORT ORDER

Jira rank or a numeric field, per column

CUSTOM BOARD CONFIG

Statuses dragged into columns and lanes

02 vi-dashboard

Personal Operations Dashboard

Self-hosted life dashboard: vault, health, calendar, budget on phone, desktop, wall kiosk.

HOW I USED AI

Scoped the modules and the three form factors up front, then built each surface with Claude in sequence against one shared layout contract: phone, desktop and wall kiosk from the same code.

Desktop: projects and tasks, with the day's focus block at the top
Wall kiosk: the day readable from across the room, with focus block, today's list, health, habits
Watch face: enhanced time awareness. Pic taken from android dev environment.
Health: sleep stages, HRV, stress and weight, with a banner when watch capture stops
Phone: the same data, plus capture and chat
Air: particulate charted three ways. Local fires cause spikes in poor air quality.
Weather: a local shed sensor with a fallback source, and three months of alert history
The agent reports back in plain speech

Digital assistant, powered by Claude

VOICE INSTRUCTION

“add 7g sugar for smoothie and record $86.84 for entertainment”

01 Read habit-logging.md
02 Glob budget/*.csv
03 Read 2026-07.csv
04 Write 2026-08.csv
BUILT ON

Claude Agent SDK, Whisperflow and the ElevenLabs API: a voice-interactable virtual intelligence that takes actions across my data and reports back

03 mtgdb

Magic Collection Manager

Self-hosted Magic collection browser and deckbuilder on a local Scryfall mirror.

HOW I USED AI

Claude built the Scryfall ingest and the search layer over 12,050 cards. I drove the schema, the price and legality logic, and the collection-matching pass the deckbuilder runs against.

The collection browser over a 12,050-card local Scryfall mirror
Deckbuilder: 99 cards, live prices and format legality, matched against what I already own
Canvas: the same deck grouped by the role each card plays
Price movers: 30-day gainers across the collection, losers across the wishlist
04 game-one

Strategy Role Playing Game

2D fantasy strategy game in Python/pygame: isometric overworld plus tactical battles.

HOW I USED AI

To answer the question: "Can I make a full-featured video game with Claude Code?" I designed the engine from scratch, one system at a time including, isometric projection, pathfinding, then the battle resolver, keeping it playable at every step.

The isometric overworld: settlements, terrain and army movement
Tactical real-time battles: control your generals and their special abilities, set troop formations, and issue battle orders
Strategic gameplay: claim castles, place stewards, increase troop reserves, manage stationed generals
Battle simulator: a standalone harness for tuning class and troop stats and game balance