projects.json

{
  "profile": {
    "name": "JS",
    "handle": "00xJS",
    "stats": [
      {"value": "2,500+", "label": "contributions in the last year"}
    ],
    "links": {"github": "https://github.com/00xJS"}
  },
  "constellations": [
    {
      "id": "fieldwork",
      "name": "At Work",
      "subtitle": "The day job",
      "color": "#f2a65a",
      "blurb": "The day job: automations, analyses, maps and data audits that keep a food bank's partner network accurate and help feed two-thirds of Arizona."
    },
    {
      "id": "stilwater",
      "name": "Saints Row Modding",
      "subtitle": "Reverse engineering & mods",
      "color": "#c79bff",
      "blurb": "The Saints Row (2006) engine, reverse-engineered into 48 research reports — then a map editor, a character creator, mods and a Mac port."
    },
    {
      "id": "ledger",
      "name": "Markets & Finance",
      "subtitle": "Backtests, dashboards & calculators",
      "color": "#f5d061",
      "blurb": "Questions about money answered with real history instead of hunches."
    },
    {
      "id": "trainer",
      "name": "Pokémon GO",
      "subtitle": "Community & data",
      "color": "#5ec8ff",
      "blurb": "Running a local Pokémon GO community turned into data projects: dashboards, maps, generators and a Discord relay."
    },
    {
      "id": "apps",
      "name": "Apps",
      "subtitle": "iPhone & web apps",
      "color": "#ff7eb6",
      "blurb": "Full apps that run on an iPhone: a household planner shipped in five weeks, and a 4X strategy game against 249 AI commanders."
    },
    {
      "id": "tools",
      "name": "Tools & Experiments",
      "subtitle": "Utilities, ports & hacks",
      "color": "#6ee7a8",
      "blurb": "Small tools and deep dives: an image combiner, a browser extension, old games ported to Apple Silicon, a PSP audio streamer and an early data notebook."
    }
  ],
  "projects": [
    {
      "id": "azfoodhelp",
      "name": "AZFoodHelp Inquiry Automation",
      "constellation": "fieldwork",
      "kind": "Automation + dashboard",
      "work": true,
      "description": "A 6-phase Power Automate system that routes every AZFoodHelp.org inquiry to the right team, logs it, and reports on it in a live Power BI dashboard.",
      "stack": ["Power Automate", "SharePoint", "Excel", "Power BI"],
      "flagship": true,
      "caseStudy": {
        "why": "Contact-form emails for AZFoodHelp.org were sorted by hand. The goal: route every inquiry to the right team automatically, measure how fast each gets answered, and show it all live — without a minute of downtime while it was built.",
        "highlights": [
          "Keyword classifier built from 897 historical submissions, routing 10 categories (Home Delivery, Volunteers, Food Donations, Finance, Partnerships…)",
          "Three flow versions iterated with the previous one still live, so routing never went down",
          "12-column SharePoint intake log: Arizona-time timestamps, response times, and a quality flag with manual override",
          "Power BI dashboard with 5 KPI cards and 8× daily scheduled refresh — manual email handling dropped to near zero"
        ],
        "learned": "Deeply nested conditions broke the flow; flattening them to run in sequence made it both reliable and easier to explain to stakeholders."
      }
    },
    {
      "id": "grocery-rescue",
      "name": "Grocery Rescue Network Map",
      "constellation": "fieldwork",
      "kind": "GIS + grant support",
      "work": true,
      "description": "A multi-layer ArcGIS map of agencies, donors and grocery-rescue partners across 76,000 square miles — and a visual case for grants: who can store food, who can't, and where capacity would grow the network.",
      "stack": ["ArcGIS", "Excel", "Grant support"],
      "caseStudy": {
        "why": "The food resourcing team needs one picture for two jobs: deciding where to source food, and showing funders why a grant matters. That means seeing where partners are, what each can store, and where the need is that no partner reaches yet.",
        "highlights": [
          "Layers agency locations, food donor sites, grocery-rescue partners and Feeding America opportunity data",
          "Shows which partners have cold, frozen or dry storage — and which have none — so a grant request can point to specific gaps",
          "Heat map of zip-code-level web inquiries over agency coverage, surfacing high-need areas with no nearby service",
          "Turns 'where should capacity go next?' into a map the team can put in front of funders and leadership"
        ],
        "learned": "A map turns 'we think there's a gap' into a place a funder can point at."
      }
    },
    {
      "id": "shutdown-map",
      "name": "Federal Shutdown Impact Map",
      "constellation": "fieldwork",
      "kind": "GIS analysis",
      "work": true,
      "description": "Built during the federal government shutdown: an ArcGIS map of where affected federal workers are, how many at each location, and the closest agency partners — so the food bank knew where to send additional resources.",
      "stack": ["ArcGIS", "Geospatial analysis", "Emergency response"],
      "caseStudy": {
        "why": "When the federal government shut down, federal workers went without pay and the food bank needed to know where that need would land. The question: who is affected, where are they, and which agency partners are closest?",
        "highlights": [
          "Mapped the federal locations whose staff would be affected by the shutdown",
          "Showed how many people work at each location",
          "Layered in the closest agency partners to each site",
          "Pointed to where additional resources needed to go"
        ],
        "learned": "In a fast-moving crisis, a map answers 'where do we send help?' faster than any spreadsheet."
      }
    },
    {
      "id": "erp-audit",
      "name": "ERP Migration Audit",
      "constellation": "fieldwork",
      "kind": "Data migration audit",
      "work": true,
      "description": "Owned the agency data sheet for an ERP migration end to end: built the document, pulled the data, made sure every field was filled and captured, and fielded the questions.",
      "stack": ["Excel", "Data quality", "Stakeholder support"],
      "caseStudy": {
        "why": "An ERP migration is only as good as the data that goes into it. The agency records needed one clean, complete, agreed structure before user acceptance testing — and someone to own it from template to sign-off.",
        "highlights": [
          "Created the document: worked with stakeholders to identify every required column and structured it to management's spec",
          "Pulled the data: researched and populated values from existing agency records",
          "Checked completeness: audited the sheet so every required field was filled and captured before UAT",
          "Fielded questions: point of contact after distribution for field values, data expectations and entry rules"
        ],
        "learned": "Most migration risk is in definitions, not data — agreeing what a field means comes first."
      }
    },
    {
      "id": "hhp-analysis",
      "name": "Households-to-Pounds Analysis",
      "constellation": "fieldwork",
      "kind": "Analysis + dashboard",
      "work": true,
      "description": "Are agency partners overserving, underserving or just right? Stitched a year of distribution reports and a full agency poundage report into one dataset with an agency ID key, then built an interactive dashboard for stakeholders.",
      "stack": ["Report consolidation", "Data modeling", "Interactive dashboard"],
      "caseStudy": {
        "why": "Pounds distributed only mean something next to the households they reach. The question was whether each agency partner is overserving, underserving or just right — and the answer was spread across a year of distribution reports plus one very large poundage report.",
        "highlights": [
          "Stitched 24 reports from across the year tracking agency distributions",
          "Joined them with a large report of all agency poundage",
          "Built an agency ID key so every partner lines up from report to report",
          "Compared households served to pounds received to classify each partner as overserving, underserving or just right",
          "Interactive dashboard so stakeholders can explore the results themselves"
        ],
        "learned": "The agency ID key was the unglamorous part — and the part that made every report joinable."
      }
    },
    {
      "id": "l2f-audit",
      "name": "Link2Feed System Audit",
      "constellation": "fieldwork",
      "kind": "Data reconciliation",
      "work": true,
      "description": "After the ERP switch, the Link2Feed neighbor database still carried old-system IDs, addresses and contacts. Reconciled Link2Feed against NetSuite and updated every agency card in the network by hand.",
      "stack": ["Link2Feed", "NetSuite", "Data reconciliation"],
      "caseStudy": {
        "why": "Agency partners log every neighbor visit in Link2Feed. After the move to the new ERP, its agency records still matched the old system — old IDs, outdated addresses and contacts — so they no longer lined up with NetSuite.",
        "highlights": [
          "Exported all Link2Feed agency data and all NetSuite agency data",
          "Synthesized the two to find mismatched IDs, addresses and contacts",
          "Updated every agency card in the network manually so both systems agree"
        ],
        "learned": "A migration isn't finished at the ERP — every system that points at it needs cleaning up too."
      }
    },
    {
      "id": "appointment-audit",
      "name": "Appointment Plan Audit",
      "constellation": "fieldwork",
      "kind": "Operations data audit",
      "work": true,
      "description": "After the ERP switch, pallet counts on some appointment plans — but not all — had been doubled to work around a pallet-volume rounding issue, which hid what agencies actually receive. Exported every plan, worked through each with its rep and Transportation, then imported 1,200+ corrections.",
      "stack": ["NetSuite", "Data cleanup", "Stakeholder sign-off"],
      "caseStudy": {
        "why": "The new ERP rounded pallet volumes in a way that didn't match real distributions, and doubling pallet counts on some plans had been tried as a workaround. Because only some plans were changed, there was no blanket fix — and the actual distribution amounts for each agency were no longer clear.",
        "highlights": [
          "Exported every appointment plan in the system",
          "Worked out which plans had doubled counts and which didn't — a plan-by-plan check, not a blanket undo",
          "Sat with each rep to understand their caseload",
          "Met with Transportation to approve and confirm the changes",
          "Imported 1,200+ updates manually"
        ],
        "learned": "Fixing data that several teams depend on takes agreement from each of them — the corrections only stuck because every rep and Transportation signed off."
      }
    },
    {
      "id": "vivery-audit",
      "name": "Vivery Data Audit",
      "constellation": "fieldwork",
      "kind": "Directory data audit",
      "work": true,
      "description": "Vivery powers the find-food-near-you search on AZFoodHelp, but no one was managing its accuracy. Pulled all of it through the API and audited every organization, location and program in a three-level hierarchy.",
      "stack": ["Vivery API", "Data audit", "Hierarchy modeling"],
      "caseStudy": {
        "why": "Neighbors use Vivery on AZFoodHelp to find the closest food pantry instead of hunting for one. The system had been left undermanaged — no one owned updates or accuracy.",
        "highlights": [
          "Exported all system data through an API pull",
          "Built a three-level hierarchy spreadsheet to audit every organization, location and program",
          "Corrected stale records and added new agency sites across the 76,000-square-mile service area"
        ],
        "learned": "A public directory needs an owner — accuracy starts slipping the moment nobody's checking."
      }
    },
    {
      "id": "sr1-map-studio",
      "name": "SR1 Map Studio",
      "constellation": "stilwater",
      "kind": "Web app",
      "url": "https://sr1-mapstudio.netlify.app/",
      "repo": "sr1-map-studio-web",
      "image": "images/sr1-map-studio.webp",
      "started": "2026-10-22",
      "flagship": true,
      "description": "Build Saints Row 1 multiplayer maps in the browser, then press Install in game and play them in Saints Reborn.",
      "stack": ["TypeScript", "Three.js", "Reverse engineering"],
      "caseStudy": {
        "why": "Saints Row (2006) never had a map editor. Map Studio is one — in the browser, with nothing to install — and it exports straight into the running game.",
        "highlights": [
          "Map wizard: 9 settings × 10 layouts, previewed from above, always fair for both teams",
          "57 ready-made maps sized like the retail ones, and ~670 placeable pieces",
          "Walk-in rooms, real textures, photo-to-building and graffiti tagging",
          "Built on 48 research reports into the engine's formats"
        ],
        "learned": "Reverse engineering is the slow part; once the formats are understood, a friendly editor is the fun part."
      }
    },
    {
      "id": "sr1-creator",
      "name": "SR1 Creator",
      "constellation": "stilwater",
      "kind": "Web app",
      "url": "https://sr1-playercreator.netlify.app/",
      "repo": "SR1-Creator",
      "image": "images/sr1-creator.jpg",
      "started": "2026-10-22",
      "description": "The Saints Row character creator rebuilt in the browser with the game's real meshes and morphs — plus photo-to-character fitting.",
      "stack": ["Three.js", "WebGL", "MediaPipe"],
      "caseStudy": {
        "why": "The game's character creator, outside the game — so you can design and share a character, or fit one to photos, then enter the values in the real menu.",
        "highlights": [
          "Parses the game's packfiles, Xbox 360 textures, meshes and morph targets in the browser — ships no game data",
          "Skin, hair and eye shaders ported from the game's own pixel shaders",
          "Photo fit: 478 face landmarks per photo solve one set of face sliders across several photos"
        ],
        "learned": "Matching a 2006 renderer means mirroring its quirks exactly — including its load-time fix-ups."
      }
    },
    {
      "id": "sr-save-editor",
      "name": "SR Save Editor",
      "constellation": "stilwater",
      "kind": "Desktop app",
      "private": true,
      "started": "2026-10-04",
      "description": "Save editor for Saints Reborn on Mac and Windows: cash, weapons, wardrobe, garage, neighbourhoods, homies and the online character.",
      "stack": ["C#", "Mac + Windows"],
      "caseStudy": {
        "why": "A save editor is the friendliest mod there is — change your story save without touching game files.",
        "highlights": [
          "Story save: cash and respect, weapon wheel, wardrobe, garage, crib cache, neighbourhoods and homies",
          "Online character: cash, owned clothing and outfit",
          "Same features on Mac and Windows"
        ]
      }
    },
    {
      "id": "sr-tweak-menu",
      "name": "SR Tweak Menu",
      "constellation": "stilwater",
      "kind": "Game mod",
      "repo": "SR-Tweak-Menu",
      "started": "2026-10-09",
      "description": "An in-game cheat menu for Saints Reborn: god mode, wanted level, rocket ammo, vehicle and people spawner, bodyguards, emotes, free camera and more.",
      "stack": ["Lua", "Keyboard + controller"],
      "caseStudy": {
        "why": "Every open-world game deserves a trainer you can drive with a controller.",
        "highlights": [
          "Opens on F5, or RB + D-pad Down on a controller",
          "God mode, wanted level, rocket and fire ammo, vehicle and people spawner, bodyguards, emotes",
          "Works on the Windows and Mac builds"
        ]
      }
    },
    {
      "id": "sr-teleport",
      "name": "Teleport to Waypoint",
      "constellation": "stilwater",
      "kind": "Game mod",
      "repo": "SR-Teleport-to-Waypoint",
      "started": "2026-10-10",
      "description": "Drop a waypoint on the map and you're there — on the rooftop, with your car and homies. No more driving across Stilwater.",
      "stack": ["Lua", "C++ add-on"],
      "caseStudy": {
        "why": "The fun part of Stilwater is rarely the drive to it.",
        "highlights": [
          "Lands on the highest surface at the spot — roads, rooftops, the top of the Ultor dome",
          "Your car and homies come with you",
          "Only your own green waypoint triggers it, never a mission marker"
        ]
      }
    },
    {
      "id": "sr-more-homies",
      "name": "More Homies",
      "constellation": "stilwater",
      "kind": "Game mod",
      "repo": "SR-More-Homies-Mod",
      "image": "images/more-homies.webp",
      "started": "2026-10-09",
      "description": "Recruit up to 10 homies instead of 3, with all ten heads on the HUD — and bigger gang wars to use them in.",
      "stack": ["Lua", "INI settings"],
      "caseStudy": {
        "why": "The game's party list already held ten followers; the game just never let you fill it.",
        "highlights": ["Keeps the game's own progression, just with a higher cap", "HUD shows all ten follower heads", "Experimental entourage mode for up to 30"]
      }
    },
    {
      "id": "saintsreborn-mac",
      "name": "Saints Reborn on Mac",
      "constellation": "stilwater",
      "kind": "Mac port",
      "repo": "SaintsReborn-Mac",
      "started": "2026-10-04",
      "description": "Play Saints Row (Xbox 360, 2006) on Apple Silicon Macs: a double-click installer that builds the game from your own disc, plus a mod launcher app.",
      "stack": ["C++", "macOS", "Apple Silicon"],
      "caseStudy": {
        "why": "Saints Reborn brought the 360 game to Windows. This brings it to the Mac — free, from your own disc, with no game files in the repo.",
        "highlights": [
          "Double-click installer that downloads the free tools and builds the game",
          "Launcher app that opens the mod loader — tick mods, press Play",
          "GPU and frame-rate tuning, with fixes merged back upstream"
        ]
      }
    },
    {
      "id": "chronoticker",
      "name": "Chronoticker",
      "constellation": "ledger",
      "kind": "Web app",
      "url": "https://chronoticker.netlify.app/",
      "repo": "Chronoticker",
      "started": "2026-04-28",
      "flagship": true,
      "description": "Stock portfolio backtester — simulate any allocation against the S&P 500 with rebalancing & DCA.",
      "stack": ["Chart.js", "Yahoo Finance", "Vanilla JS"],
      "caseStudy": {
        "why": "Would that allocation have beaten the S&P 500? Chronoticker replays portfolio allocations against real history — with rebalancing and dollar-cost averaging — so the answer is a simulation, not a hunch.",
        "highlights": [
          "Backtests any allocation against the S&P 500",
          "Models rebalancing strategies and DCA schedules",
          "Charts the full growth path, not just the end number"
        ],
        "learned": "Financial simulations live or die on edge cases — contribution timing and rebalance rules matter more than they look."
      }
    },
    {
      "id": "blockwatch",
      "name": "Blockwatch",
      "constellation": "ledger",
      "kind": "Live dashboard",
      "url": "https://btcblockwatch.netlify.app/",
      "repo": "Blockwatch",
      "started": "2026-04-28",
      "description": "Live Bitcoin on-chain dashboard — market, network health, and sentiment, plus comparisons to gold and the national debt.",
      "stack": ["Chart.js", "CoinGecko API", "mempool.space"],
      "caseStudy": {
        "why": "One screen for the Bitcoin questions that used to take five tabs: price, network health, sentiment, and how it all stacks up against gold and the national debt.",
        "highlights": [
          "Live market and on-chain data from keyless public APIs (CoinGecko, mempool.space)",
          "Comparisons that give the numbers scale — gold, the US national debt",
          "Fails gracefully when any upstream API is down"
        ],
        "learned": "Building resilient dashboards on free, keyless APIs — caching, fallbacks, and honest handling when upstream data goes stale."
      }
    },
    {
      "id": "cashflow-comp",
      "name": "Cashflow Compass",
      "constellation": "ledger",
      "kind": "Web app",
      "url": "https://cashflow-comp.netlify.app/",
      "repo": "cashflow-compass",
      "started": "2026-04-26",
      "description": "See your financial future month by month, up to 3 years ahead — browser-only, with Excel round-trips.",
      "stack": ["Chart.js", "SheetJS", "Modular JS"],
      "caseStudy": {
        "why": "Spreadsheets can model your financial future, but they don’t show it. Cashflow Compass projects cashflow month by month, up to three years out — and keeps Excel round-trips working so nobody has to abandon their budget.",
        "highlights": [
          "36-month projection engine running entirely in the browser",
          "Excel import and export via SheetJS — meet budgets where they already live",
          "Projection engine kept separate from the UI, so features land cleanly"
        ],
        "learned": "Separating the engine from the interface paid for itself with every feature added afterward."
      }
    },
    {
      "id": "project401k",
      "name": "Project 401k",
      "constellation": "ledger",
      "kind": "Web app",
      "url": "https://project401k.netlify.app/",
      "repo": "project-401k",
      "started": "2026-04-06",
      "description": "401(k)-style dollar cost averaging into the S&P 500 — SPY closes, employer match, dividends reinvested.",
      "stack": ["Chart.js", "Vanilla JS"],
      "caseStudy": {
        "why": "A payroll contribution is the most common investment plan there is, and the hardest to picture. Project 401k replays one against SPY's real closes — your contribution, the employer match, dividends reinvested — and shows what the account would actually be worth.",
        "highlights": [
          "Replays contribution schedules against SPY closes back to 1993",
          "Counts the employer match separately from your own money",
          "Reinvests dividends and shows the shares they bought"
        ],
        "learned": "Leaving dividends out understated a 26-year plan by $165,790 — the honest version needed the data, not a footnote."
      }
    },
    {
      "id": "calculat-bur",
      "name": "Calculat-Bur",
      "constellation": "ledger",
      "kind": "Calculator",
      "url": "https://calculat-bur.netlify.app/",
      "repo": "calculat-bur",
      "started": "2026-04-08",
      "description": "Convert BTC to USD at live prices, then re-price that holding from $50K to $1M — arithmetic, not forecasts.",
      "stack": ["Chart.js", "CoinGecko API"],
      "caseStudy": {
        "why": "A conversation-settler: what is this BTC holding worth right now, and what would it be worth at $50K to $1M per coin? Arithmetic at live prices — deliberately not forecasts.",
        "highlights": [
          "Live BTC pricing via the keyless CoinGecko API",
          "Re-prices a holding across a $50K–$1M range instantly",
          "Forecast-free on purpose: it’s a calculator, not a crystal ball"
        ],
        "learned": "Scoping a tool down to one honest job makes it more trustworthy — and somehow more useful."
      }
    },
    {
      "id": "dca",
      "name": "DCA Analysis",
      "constellation": "ledger",
      "kind": "Web app",
      "url": "https://dca-btc-with-me.netlify.app/",
      "repo": "dca-analysis",
      "started": "2025-02-11",
      "description": "Dollar Cost Averaging, replayed on real Bitcoin prices — drag the start date and watch the average cost move.",
      "stack": ["Chart.js", "Vanilla JS"],
      "caseStudy": {
        "why": "Dollar-cost averaging is easy to preach and hard to picture. DCA Analysis replays a real DCA plan against Bitcoin's own price history and shows exactly what would have happened.",
        "highlights": [
          "Replays DCA schedules against real historical prices",
          "Drag the start date and watch the average cost move",
          "Honest accounting: totals, cost basis, and the rough patches too"
        ],
        "learned": "A replay of history persuades people in a way projections never do."
      }
    },
    {
      "id": "pogo-metrics",
      "name": "POGO Metrics",
      "constellation": "trainer",
      "kind": "Web app",
      "url": "https://pogo-metrics.netlify.app/",
      "repo": "POGO-Metrics",
      "started": "2026-08-02",
      "flagship": true,
      "description": "Turn your official Niantic Pokémon GO data export into a private, in-browser dashboard of your trainer journey.",
      "stack": ["Vanilla JS", "In-browser parsing", "SVG"],
      "caseStudy": {
        "why": "Niantic will email you every scrap of data Pokémon GO holds on you — as a pile of raw files nobody can read. POGO Metrics turns that export into a dashboard of your trainer journey, without the data ever leaving your browser.",
        "highlights": [
          "Parses the official Niantic data export entirely client-side — nothing is uploaded",
          "Distills years of gameplay into charts you can actually explore",
          "Zero backend: static hosting, no accounts, no tracking"
        ],
        "learned": "Privacy as a feature — processing a sensitive export in the browser removes the scariest part of every “upload your data” tool."
      }
    },
    {
      "id": "abc-analysis",
      "hidden": true,
      "name": "ABC Applied Analysis",
      "constellation": "trainer",
      "kind": "ASU capstone",
      "private": true,
      "started": "2025-02-10",
      "description": "ASU analytics capstone: building Buckeye's Pokémon GO community from scratch, then measuring which events and formats actually brought players out.",
      "stack": ["Excel", "Data viz", "Campfire check-ins"],
      "caseStudy": {
        "why": "Buckeye, Arizona had no organized Pokémon GO community. The capstone for ASU's Applied Business Data Analytics certificate was to build one — and use the check-in data to find out what worked.",
        "highlights": [
          "Scheduled events as soon as they were announced instead of 1–3 days out, so families could plan",
          "Standardized event posts (via Sundance DeGen) and a single central raid meeting point",
          "Check-in analysis found Gigantamax events drew the most players — 31 on average"
        ],
        "learned": "Community building is an operations problem: consistency and lead time beat hype."
      }
    },
    {
      "id": "trainer-model",
      "name": "POGO Trainer Model",
      "constellation": "trainer",
      "kind": "Data study",
      "url": "https://pogo-metrics.netlify.app/trainer-model.html",
      "repo": "pogo-trainer-model",
      "started": "2025-02-08",
      "description": "Started as a linear regression in a notebook; became an interactive study of 883 trainers across two eras of the level cap — including where the first model went wrong.",
      "stack": ["Regression", "Jupyter → JS", "Data viz"],
      "caseStudy": {
        "why": "Can trainer level predict battles won, distance walked and Pokémon caught? One friends list, recorded under the old level-50 cap and again under the new cap of 80, gave a real dataset to find out.",
        "highlights": [
          "390 trainers under the cap-50 era, 493 under cap-80",
          "Explorer to switch fits (linear, log-linear, median) and outlier handling and watch the model move",
          "A 'model report card' grading the original regression against alternatives — and reality"
        ],
        "learned": "The level-50 wall capped every level-based model; admitting where the first model failed became the most interesting result."
      }
    },
    {
      "id": "silph-relay",
      "name": "Silph Relay",
      "constellation": "trainer",
      "kind": "Automation",
      "url": "https://silph-relay.netlify.app/",
      "repo": "silph-relay",
      "started": "2026-03-25",
      "description": "A $0/month Discord relay — six Pokémon X accounts pushed to two channels in under a minute, with an analytics dashboard.",
      "stack": ["Python", "GitHub Actions", "Discord webhooks"],
      "caseStudy": {
        "why": "Six Pokémon accounts post into a Discord relay that costs $0/month to run — no X API, no server. The dashboard answers the operational questions: how often does each feed post, and how fast does a post land?",
        "highlights": [
          "GitHub Actions plus a free cron trigger — no server, no paid API",
          "Tracks post frequency and delivery latency per feed",
          "Static analytics dashboard on the same $0 footprint"
        ],
        "learned": "Even a hobby automation deserves observability — measuring latency changed how the relay was tuned."
      }
    },
    {
      "id": "degen",
      "name": "Sundance DeGen",
      "constellation": "trainer",
      "kind": "Generator",
      "url": "https://sundance-dgen.netlify.app/",
      "repo": "sundance-dgen",
      "started": "2025-01-28",
      "description": "Turn a month's Pokémon GO schedule into copy-ready Campfire meetup posts — artwork, catch CP, and shiny status filled in.",
      "stack": ["Vanilla JS", "Templating"],
      "caseStudy": {
        "why": "Writing Campfire meetup posts for a month of Pokémon GO events is repetitive community work. DeGen turns the schedule into copy-ready posts — artwork, catch CP, and shiny status filled in automatically.",
        "highlights": [
          "Generates a month of meetup posts from a markdown schedule",
          "Auto-fills artwork, catch CP, and shiny availability",
          "Output formatted and ready to paste into Campfire"
        ],
        "learned": "Automating the boring 20% of community organizing keeps the fun 80% sustainable."
      }
    },
    {
      "id": "mickeys",
      "name": "MickeyDeez Map",
      "constellation": "trainer",
      "kind": "Map",
      "url": "https://phoenix-mickydeez.netlify.app/",
      "repo": "phoenix-mickeydeez",
      "started": "2025-01-23",
      "description": "A crowdsourced map of Phoenix's Pokémon GO–sponsored McDonald's — 31 stops, each verified as an active raid Gym.",
      "stack": ["Python", "Folium", "Leaflet"],
      "caseStudy": {
        "why": "Phoenix’s McDonald’s-sponsored PokéStops were scattered folklore. This map makes them a verified dataset: 31 stops, each confirmed as an active raid Gym.",
        "highlights": [
          "Interactive Leaflet map of all 31 sponsored locations",
          "Every stop community-verified as an active raid Gym",
          "Geocoded for route planning, built with Python and Folium"
        ],
        "learned": "A small, verified dataset beats a big, stale one — curation is the product."
      }
    },
    {
      "id": "novacore",
      "name": "Galaxy Royale",
      "constellation": "apps",
      "kind": "iPhone game",
      "repo": "novacore",
      "image": "images/novacore.webp",
      "imageFit": "portrait",
      "started": "2026-07-30",
      "flagship": true,
      "description": "A mobile 4X where one human fights 249 simulated commanders in a 2500×2500-tile galaxy that keeps living while you sleep.",
      "stack": ["Unity 6", "C#", "Supabase"],
      "caseStudy": {
        "why": "A research question: could a mobile 4X go from a blank folder to running on my own iPhone, designed and redesigned conversationally with an AI coding assistant? The answer was yes — three times over.",
        "highlights": [
          "249 AI commanders that play by the same rules you do",
          "One persistent galaxy that simulates a live MMO server, single-player",
          "Three incarnations: a TypeScript prototype, an online Unity build, and this final form"
        ],
        "learned": "Started as Novacore and became Galaxy Royale — the rename is part of the story. Rough edges are part of the artifact."
      }
    },
    {
      "id": "smalltown-hustle",
      "hidden": true,
      "name": "Smalltown Hustle",
      "constellation": "apps",
      "kind": "Browser game",
      "private": true,
      "started": "2026-10-08",
      "description": "A life-sim RPG in the spirit of the Flash era: one small city, a 24-hour day, and ambitions honest work can't afford.",
      "stack": ["Phaser 4", "TypeScript", "Vite"],
      "caseStudy": {
        "why": "An original take on the Flash-era life RPGs, built so the whole economy can be tuned without touching code.",
        "highlights": [
          "Every tunable number — jobs, prices, rates — lives in one economy file",
          "Seeded RNG that rolls at sleep, so reloading can't reroll the day",
          "Work, study, gamble or smuggle your way to one of several endings"
        ]
      }
    },
    {
      "id": "food-banker",
      "hidden": true,
      "name": "The Food Banker",
      "constellation": "apps",
      "kind": "iPhone game",
      "private": true,
      "started": "2026-09-30",
      "description": "An operations sim about how a food bank moves food, money, people and data — and how it grows from one warehouse to a network.",
      "stack": ["Unity 6", "C#", "iOS"],
      "caseStudy": {
        "why": "The day job, as a game: a cutaway food bank where every room only works with volunteers in it.",
        "highlights": [
          "Rooms from the dock to the pickup lane that run continuously",
          "Grows from a warehouse to a distribution center, storefront and rural sites",
          "Seasons and a yearly challenge in its legacy years"
        ]
      }
    },
    {
      "id": "drafter",
      "name": "Drafter",
      "constellation": "apps",
      "kind": "Web + iPhone app",
      "private": true,
      "started": "2026-08-29",
      "flagship": true,
      "description": "A household planner and journal — tasks, calendar, money, meals, people and a daily journal — shipped for web and iPhone in five weeks.",
      "stack": ["TypeScript", "Supabase", "iOS", "MCP"],
      "caseStudy": {
        "why": "One app for running a home: tasks, calendar, money, meals, people, places, wardrobe and a journal — offline-first, shared with the household.",
        "highlights": [
          "Web and iPhone app shipped in five weeks",
          "MCP server with ~47 tools, so AI assistants can work with your records — with the permissions you give them",
          "LLM-powered Q&A over personal records; Google and Outlook calendar sync"
        ],
        "learned": "Local-first sync needs an honest conflict story — 'Keep mine' beats silently losing an edit."
      }
    },
    {
      "id": "remotejoy-audio",
      "name": "RemoteJoy Audio",
      "constellation": "tools",
      "kind": "Homebrew",
      "repo": "remotejoy-audio",
      "started": "2026-08-19",
      "flagship": true,
      "description": "Stream a real PSP's screen and game audio to a Mac over one USB cable — the first PSP screen-streamer to carry audio.",
      "stack": ["C", "PSP kernel plugin", "libretro"],
      "caseStudy": {
        "why": "Every PSP streaming tool carried video only; the standard advice was a 3.5 mm cable into a capture card. This captures the audio inside the PSP and sends it down the same USB cable.",
        "highlights": [
          "Kernel plugin hooks the PSP's audio output calls and mixes all 8 hardware channels at the right levels",
          "Video and audio multiplexed over USB bulk transfers",
          "libretro client in RetroArch, native on Apple Silicon"
        ],
        "learned": "Capturing one channel gives you part of a soundtrack — mixing every channel on a real-time timeline is what makes it sound right."
      }
    },
    {
      "id": "empire-earth",
      "name": "Empire Earth on macOS",
      "constellation": "tools",
      "kind": "Mac port",
      "repo": "Empire-Earth",
      "started": "2026-09-17",
      "description": "Run your own GOG copy of Empire Earth Gold on an Apple Silicon Mac, full screen, with free tools — and armies of up to 10,000 units.",
      "stack": ["C", "Wine", "DXVK + MoltenVK"],
      "caseStudy": {
        "why": "A 2001 classic, playable on a modern Mac without CrossOver or Parallels.",
        "highlights": [
          "Full screen at native resolution, Cmd+Tab, and notch-aware layouts",
          "~130 FPS in thousand-unit battles on an M2 Pro, with game maths moved to modern CPU instructions",
          "Unit limit raised to 10,000, with drag-select that takes the whole box"
        ]
      }
    },
    {
      "id": "icombo",
      "name": "I-Combinator",
      "constellation": "tools",
      "kind": "Web app",
      "url": "https://i-combinator.netlify.app/",
      "repo": "I-Combinator",
      "started": "2025-07-17",
      "description": "Combine images side by side or stacked, without cropping — drag, drop, reorder, and export in your browser.",
      "stack": ["Canvas API", "Drag & drop", "Vanilla JS"],
      "caseStudy": {
        "why": "Combining screenshots usually means opening an image editor for a ten-second job. I-Combinator does side-by-side or stacked combos in the browser — drag, drop, reorder, export.",
        "highlights": [
          "Canvas-based compositing with no cropping surprises",
          "Drag-and-drop ordering with instant preview",
          "Everything stays local — images never leave the machine"
        ],
        "learned": "The Canvas API is enough for a real image tool; the hard part is making layout behavior predictable."
      }
    },
    {
      "id": "edge-as-chrome",
      "name": "Edge as Chrome",
      "constellation": "tools",
      "kind": "Browser extension",
      "repo": "edge-as-chrome",
      "started": "2026-10-01",
      "description": "An Edge extension that makes Edge identify as Chrome on the sites you choose, so 'unsupported browser' messages go away.",
      "stack": ["JavaScript", "Manifest V3"],
      "caseStudy": {
        "why": "Some sites refuse Edge for no reason other than its name.",
        "highlights": [
          "Rewrites the User-Agent header, Sec-CH-UA client hints and navigator values",
          "Per-site: every other site still sees normal Edge",
          "One click to turn it on for the current site"
        ]
      }
    },
    {
      "id": "breadcrumbs",
      "name": "HomeTech Offer Analysis",
      "constellation": "tools",
      "kind": "Notebook",
      "repo": "project-breadcrumbs",
      "started": "2025-02-08",
      "description": "Exploratory analysis of customer device ownership — and which probing questions actually lead to protection-plan sales.",
      "stack": ["Python", "Pandas", "Jupyter"],
      "caseStudy": {
        "why": "Does the device a customer asks about predict the other devices they own — and which questions turn that into a protection-plan sale?",
        "highlights": [
          "EDA on a synthetic customer dataset",
          "Device-ownership profiles, from customers with many devices to those with none",
          "Compared probing questions by conversion rate"
        ]
      }
    }
  ]
}