#5: Final Tool

A problem I think a lot of people face with job applications is actually putting their resume stuff into words in an interview. Even if you have all the qualifications, you can’t exactly list them bullet-point style when face-to-face with someone. The new Coach feature solves this issue with direct integration between Resumight and Tailor. Basically, it tells you how you can best present your unique skills to the hiring team.

The End

With that, I believe the crux of the project is complete. There’s obviously more stuff I could add, and I still haven’t figured out how to speed up the AI integration, but I’ve also noticed that I’m reaching the end of the line for what I actually know how to make, and have been relying on AI Studio suggestions a bit more.

#4: Back Up

It’s been a while since I’ve made changes to Resumight, so I want to take things slow for a bit to make sure nothing breaks. After all, I’m really proud of the tools it has so far!

Running from my problems

API keys don’t last forever, so my solution was to add a section for a backup API key that only gets used once the first one comes up empty. I couldn’t think of anything else. In case you’re wondering, since the app is hosted using Google AI Studio, I can integrate the keys without having them directly in code or anything. That way this can stay open source and have a GitHub.

Adding on

I realised that the most useful thing for me to do with Resumight would actually be to update my old resume. Adding the PDF upload feature to Resumaker was very easy, but this does require token usage again. I’m also a bit concerned about how long it took to parse my one page resume, but it does work for now.

Next

The website really feels like a one and done sort of thing right now. I want to have some sort of multi-resume management system: nothing fancy, but it would use cookies and help people stay organised. Also, voice recognition really does not like any browser other than Chrome.

#3: It lives!

I cleaned up the layout and created my Resumaker and Tailor sections. One makes your resume, and the other gears it up for any possible job listing.

AI Integration (Misery)

Resumight sort of needs AI to function. Without an AI parsing and sorting your speech, the Resumaker is a glorified document editor. Without an AI to tailor your resume to a job listing, the best I can do is match keywords

Solution

Google has an awesome free tool called AI Studio which lets you get an API key, which I was able to hook up directly to Resumight. Now, Google’s Gemini automatically sorts everything you say into where it needs to go! It’s pretty smart, too (see images). No master list necessary!

The Tailor uses AI to specifically let you know what to update about your bullet points, and then makes a downloadable version for you. It’s not just buzzwords this time, since Gemini can actually explain what it wants to delete.

Next Time

API sustainability is at a critical point.

#2: Spoken Language is Hard

It’s kind of a hassle to have to input all your information into a website every time, right? I decided to start making a speech parsing thing that directly translates microphone input to words.

I found an open-source base script online, which I was able to use to create my own speech-to-text which works with just code.

Just to see how far it would go, I even tried seeing if I could get it to recognize shorthands and list them accordingly. I found a list of popular majors and schools, and just went from there. Categorization is still messy.

Solution(?)
Yeah, I’m probably going to have to use AI integration for this part, but we’ll cross the bridge of token limits once we get there.

#1: Warm regards from BPT

Figured out the color palette for my project; a crisp white and soft green are pretty nice to look at.

This has been coded in HTML, CSS (so much CSS) and JS. An app might be more functional, but I like GitHub Pages too much for that.

This image should give you a hint as to what we’ll be moving into. Let the suspense eat away at you.

Main order of business: I’m working on getting that cool scrolling effect you see on Apple websites working now, just to make the website seem more legit.

Triple Threat

Minigame 3 is a catcher minigame, where players try to catch falling “food” items in a small bucket. Just avoid the meteorites, or you might go the way of the dodo.

Our settings also has simple but basic functionality: swap between fullscreen and windowed mode, and turn music on and off.

Speaking of music, we now have background music and special jingles for winning and losing, alongside the addition of victory and death screens. Avid gamers might recognize the tunes…

With all of this added, RaptorRush is ready to rumble!

The Guided Sections

I have recreated the base content from the WarioWare game guide:

• Title Screen
• Lives System
• Platforming Minigame
• Clicker Minigame

My game puts a prehistoric twist on it, with a much more prehistorically accurate velociraptor than in most movies.

Next, I’d like to come up with an idea for a third minigame and implement that.

Also, the settings page doesn’t really exist right now, but that’s an issue for later.

FossilLog 3: Locomotive Engineering
Now I’ve made a working train that can go between stops, and is powered by the fossil fuel the dinosaurs drop. Simply go up to the train and click it with the fuel to power it and begin the journey!

Fossil fuel is good for more than that, though. It also works as a combustible to get up close and personal when the situation calls for it. Damage is scaled based on how close you are to a target; and don’t worry, it won’t damage the user.

The idea with this fossil fuel is that players will have to choose how they use their precious fuel. Will you power your steam engine to escape from danger, or will you fight back when that danger comes knocking (or biting)?

You might also notice the addition of a main menu. This will probably be updated, perhaps with something sinister lurking underneath the water… Let me know if you have any suggestions. This is a train game, but I’m sure I can figure out some way to get aquatic danger brewing.

FossilLog 2: We have life!
I made a velociraptor that chases the player around at high speeds. I was also able to make some basic animations, including a slashing claw and an opening mouth. It even jumps when you do, or when it runs into an obstacle!

Something also happens when the dinosaur touches the death part - its body disappears and is replaced by fossil fuel. This is what will be used to fuel the train I’ll complete soon.

FossilLog 1: Starting Studio

I’m beginning with the basics of making parts and scripts.

I found a tutorial online on how to navigate Studio and set up local variables. I don’t have much experience with Lua, but alongside some video help, I was able to make a script that makes a part set a player’s health to 0 when they touch it - perfect for obbies!

The video also suggested I try and make some dice, so I found dice textures from the Studio toolbox and put two cubes high in the air. Since these always start in the same position, their results don’t change, though. I know most programming languages have a random feature, so maybe that would work.

Anyway, these are baby steps. Next time, I want to try and get a whole train working. Wish me luck…

Looks pretty sketchy…

Introducing the MagSketch app, which comes with different colours, pen sizes, and even a colour picker (I’m pretty proud of that one).

Getting the canvas to work was pretty challenging, but there are plenty of guides online that really helped speed things up.

What a drag…
I’ve updated quite a bit since the last devlog, so the windows are now draggable, closable, openable, all that jazz.
We also have some editability: the MagNotes app is up and working as a simple notepad, and the to-do list in the taskbar can be updated.
One app is a little boring, though, so I think we can step things up in the coming days. The end of WebOS 1 is in sight!

Are we sure we’re on task right now?

The bottom bar is ready, except for that to-do button in the centre. That can be made inputtable and working once I get a bit more in-depth on my JavaScript.

Right now, we have a window of content, the taskbar, and a nice screensaver background (this might be a placeholder, but I kind of like it).

Moving on to some clickability!

Welcome!
I’ve made the welcome page for MagnOS, including the basics on what it does, and a little introduction to how magnets work. That’s the basic HTML and CSS, but JavaScript (a.k.a. the hard stuff) is yet to come.

If this pans out how I hope, I think the OS can really be a cool thing to mess around with, and I might even use the code for my personal productivity!

It can fly!

I updated the Onshape and reprinted the train with five magnet holes and a new spacing pattern. This means that, at all times, there is at least three magnets’ worth of propulsion. A big improvement from the previous zero, wouldn’t you say?

You can see in the image below that the ends are elevated above the track. When hit with the launcher, it maintains its inertia to spin around at a stable height.

Printed the train, but it won’t float!

I’ve made a severe and continuous lapse in my judgement.

The train’s magnet holes are spaced the same as the track… which means each magnet of the train can rest in the spaces between the magnets on the track.

So if I launch it (using the handy popsicle stick launcher pictured on the right), it will seem to float while in motion, but then abruptly stop.

Back to the CADing board. ✌