@TejalK

Built a single-file math crossword game with 5 themed puzzles and 100 clues using vanilla HTML/CSS/JS - no build tools, no dependencies. The trickiest part was implementing keyboard navigation (arrow keys, tab-to-toggle, auto-advance) while keeping movement locked to the current word. Grid generation and clue numbering work well enough, though the data format is hand-crafted rather than auto-generated. Mobile layout is functional but needs polish; localStorage save and a timer are next on the list. Overall it’s a working v1.0 that runs anywhere you can open a browser.

So my game used to look like this dark “Element Navigator” thing (second pic). It was… fine? But honestly, it felt like a textbook diagram, not a game. I nuked it and started over. Now it’s this bright purple space vibe with an actual starfield, the rocket dead center, and clean element buttons at the bottom. Way more colorful, way more fun, and your thumbs don’t have to reach across the whole screen anymore. Old layout: cramped photo of a jet, chunky circles everywhere, zero personality. New layout: clean, bold, actually looks like something you’d want to play.

Added a step where the quiz asks which country you’re from and whether you want live job listings for your country or the existing Canadian data. If you opt in, it pulls live results for your country; if not (or if there’s nothing available), it falls back to the built-in Canadian dataset. Makes the quiz useful beyond a Canada-only audience without losing the reliable baseline for users who skip it. Also cleaned up a handful of small bugs along the way.

TODAY’S WORK :

API Integration - Indeed’s job API turned out to be sales-gated with no self-serve access, so pivoted to Adzuna (free, self-serve). Wired up live job postings per career on the detail screen, with caching and graceful loading/error states. Also added a country-selection step before the quiz so postings can be localized, plus fixed a couple of rendering bugs that came with it.

Share Result - Added a “Share Result” button that generates a downloadable PNG summary card and copies a shareable link. Answers are encoded straight into the URL, so results travel with zero backend.

Compare with a Friend - New stretch feature built on the share-link encoding. Paste a friend’s link, we decode and recompute their results, then show a side-by-side comparison with overlapping matches highlighted.

Big polish day. Replaced hardcoded pixel positions with a percentage-based layout and made the canvas actually resize instead of being stuck at 1520×730. Added keyboard navigation (arrows + Enter) and in-memory progress persistence with a “resume quiz?” prompt so closing the tab doesn’t wipe everything.

Wired yesterday’s scoring work into the detail screen - you now see a confidence badge and a line like “You matched because you value Impact + chose Science” instead of just a bare result. Also wrote day-in-the-life blurbs for all 24 careers and added a related-careers section (Software Engineer -> Data Scientist, etc.) so the quiz doesn’t dead-end at your top match.

Split the quiz out of one giant file into data.js, quiz.js, scoring.js, render.js, with a basic test harness for calculateResults(). Then fixed the actual scoring bug: careers used to win on being consistently “meh” across every question instead of nailing a couple of perfect matches. Added a bonus for 3/3 scores on 2+ questions, confidence labels (“Strong Match” / “Worth Exploring”), and tracking for which questions drove each career’s score.

Built out the results screen (ranked cards with match score, salary, outlook) and a detail view for digging into one career fully - subjects, degree, specializations, and top universities.

Wired up the actual quiz - question progression, single vs. multi-select handling, and a scoring function that totals weighted points per career and sorts for the top 5 matches. Also added a university lookup table mapped by field, so the results screen can suggest real schools per career.

Before touching any UI, put together the career list - 24 careers across Medicine, Technology, Business, Engineering, Law, Education, and a few others. Each one has salary range, outlook, recommended subjects, degree, and specializations.

Then built a scoring table: every career gets weighted values across all 5 quiz questions, so answers aren’t just matched to one “type” - they add up across multiple signals (subjects, interests, motivation, study length, work style).

Revamped the Digital Carbon Footprint Calculator from a flat, static two-panel layout into a dark dashboard-style UI: swapped light-blue panels for a navy/cyan theme, replaced plain number inputs with sliders for continuous values, and broke the results into collapsible sections (View Emission Factors, View Charts, See Tree Impact) instead of dumping everything at once. Added a transparent emissions-factor table so users can see exactly how their footprint is calculated, plus a “Tree Impact” card that translates the raw CO₂ figure into a relatable equivalent (trees needed per year to offset it) - turning the tool from a bare calculator into a clearer piece of climate communication.

Started this because I kept noticing how scattered career research actually is — subject requirements in one place, degree info in another, salary data somewhere else, university rankings somewhere else entirely. Wanted something that pulls all of that into one quick interactive tool instead of 10 open tabs.

Plan: a short quiz (5 questions) that scores your answers against a list of careers, then shows your top matches with everything you’d need to know — subjects to focus on now, degree to aim for, and where to study it.

I’d already built the core of EcoDigital Tracker before deciding to bring it into Stardance, so this session was about picking it back up and turning it into something actually shareable. It started as a desktop app built with Tkinter — you could enter your daily digital habits (emails, streaming, cloud storage, calls, browsing) and get back an estimated carbon footprint in grams of CO₂, along with a Normal/High/Severe status and a pie + bar chart breakdown. The problem was that it only ran on my machine — no link, no way for anyone else to try it without installing Python themselves.

So I rebuilt the UI in Streamlit, keeping the exact same layout, input fields, output format, and charts, just swapping the framework underneath so it could run in a browser instead of a native window. I also wrote a short README explaining what the tool does and why I built it: to show people how much carbon their everyday digital activity actually produces, and where it lands on the Normal/High/Severe scale.

From there I moved on to getting a real public link. I tried ngrok first to tunnel my local server, but the download got flagged by my antivirus, so I switched to deploying properly instead. I pushed the project to GitHub and connected it to Streamlit Community Cloud, then worked through a chain of deployment errors — a missing requirements.txt that meant matplotlib wasn’t installed on the server, a main file path pointing at a filename with no .py extension, and the old Tkinter file accidentally being the one that got deployed instead of the Streamlit version. Fixed each one by correcting the filenames and adding requirements.txt directly on GitHub.

I have added new levels (Spacecrafts) to my game and I might also introduce new domains to this game like underground, underwater, etc. So, stay tuned!!!