PROJECT
Dec 2023 • Fullstack Developer & Researcher
BRAND
We're diving into this cool research project, and our main gig is to create a super user-friendly website that's all about grading essays automatically. We want to make life easier for teachers. So, here's the deal: we're using fancy tech stuff to make grading essays a breeze. The goal? To speed things up and make sure teachers can be efficient and fair when they're checking those essays. We're all about making education better and bringing in some cool tools.
With our website, we're not just talking about saving time – we're making sure the grading game is top-notch. We're using smart algorithms to keep things on point. It's not just about being fast; it's about making sure the grading system is top-quality and fair. Our website isn't just for show – it's a game-changer for teachers drowning in essays. Quick, fair, and flawless – that's what we're aiming for. Let's give educators a tool that's efficient and keeps things real. Time to shake up the essay-grading game!
My job? Figuring out exactly what teachers need and then building a tech wonderland behind a website to speed up and level the playing field in essay evaluations. I'm digging into what teachers really need to grade essays like champs. It's all about understanding their current methods, criteria, and making sure the website is a go-to for all sorts of teachers.
So here's the lowdown on what I'm up to in this whole tech world of auto essay scoring. I'm the artist making sure the part of the website teachers mess with looks slick and is easy to handle. Teachers can throw in their essays without a hitch, get feedback in real-time, and check out their grades on a super-friendly dashboard. I'm rocking Next.js to make the front end top-notch.
I'm the brains behind the scenes, creating the logic that keeps things ticking. Managing user accounts, handling requests, and making sure the whole show runs smooth – that's my jam. Security and scalability? Big deals for me. I'm working with Nest.js using a service base to make the back end scalable and robust.
Time to bring in the AI magic! I'm using Python and a Hugging Face base model to teach the system how to grade essays all by itself. Think advanced algorithms and pre-trained models doing the heavy lifting. I'm fine-tuning the model with my data to make sure it's spot-on.
I'm the data keeper, making sure all the essays and grades are stored and found super fast. I'm juggling MariaDB and MongoDB to manage info like a boss.
I'm the messenger, setting up a system that lets all parts of the service communication smoothly. Front end, back end, and the AI crew – we're all on the same page. I'm using RabbitMQ as the go-to message broker to keep everything flowing seamlessly.
Let me tell you about this auto essay scoring gig – it's a rollercoaster! We've got a tight 4-month deadline, and the system specs are like a laundry list of features. Juggling all these tasks to stay on track is like a marathon. The tech stuff, whoa, it's a puzzle. We've got this complex structure with loads of services, and they're all holding hands like best buds. Trying to make it scalable and easy to keep in check? Not gonna lie, it's a head-scratcher.
Now, the real deal – the scoring system. Talk about a challenge! I had to fine-tune the model to match the data requirements, and that process? A resource-hungry beast. But you know what? We pushed through all these hurdles, and guess what? The results are downright thrilling!
The auto essay scoring results? They're legit impressive. I threw it into a scenario with 52 peeps answering 10 questions. Boom, it took just 7 minutes! That's less than 1 second for each response. Imagine the time we're saving teachers on the grading hustle. Checked the score results, and it's not just random numbers – we got some solid correlation going on. A sweet average of 0.7! Translation: the scores are on point, and you can trust 'em. It's like having a grading superhero on standby.
Check the demo at inkwell.my.idLOADING...
Brace yourself for epicness!
