FutureFocus | Devpost

Inspiration

The modern job market is extremely competitive with ever increasing job demands and stricter ATS software thinning thousands of online applicants down to just a few lucky candidates. Our team wanted to help young software engineers, and anyone else looking for guidance, to progress through their career with the help of a tailor made, incremental, AI generated plan.

How it works

Our application starts with the user telling the app their desired career and current experience level. They may also upload their resume to further increase the amount of detail the model can put into their plan. The website then returns a list of recommended steps they can take to further their career, each split up into three categories based on how long they might take. If or when the user successfully accomplishes one of the steps, they may check that item off the list and regenerate the it with the updated context. The result is a recursive, game-like loop, hopefully keeping the user progressing along in their career while building key resume skills.

How we built it

We decided to use SvelteKit for both the frontend and backend due to its ease of use and quick development time. We also decided to use Tailwind CSS to style our frontend for similar reasons. By leveraging SvelteKit's form actions, we were able to make secure groq API calls using the Llama3-8b model from the backend of the same project that also served our frontend to the user. Finally, to actually host our application, we configured an AWS EC2 instance with HCP Terraform, an infrastructure as code platform. By the time our team had finalized the project's tech stack, we had a single project directory and completely automatic deployment with a single git push command. One of the largest benefits we found to this approach was how quickly each member could not only learn each technology but also pull and push changes to match the fast paced environment. This meant that we were able to spend less time debugging & configuring and more time actually creating.

Challenges we ran into

While our deployment cycle was incredibly fast, our team still had to spend time learning about the technologies we wanted to use for the project. Much time was spent reading Terraform documentation to configure our deployment exactly how we wanted. Additionally, we had to learn about Groq, what it could offer us, and how to use its API. Afterwards, we spent time setting up our SvelteKit actions and determining the general flow of the app's frontend. Near the end of the app's development, we decided to implement PDF scraping and had to determine which of the available solutions would best suit our needs, as PDF files are not a standardized format.

Accomplishments that we're proud of

Streamlined development and deployment cycle
Configuring AWS with Terraform
Scraping user provided PDF files
Prompt engineering efforts

What we learned

How to use HCP Terraform, SvelteKit, and the Groq API.

What's next for FutureFocus

Better frontend styling would greatly benefit the project. Additional features such as importing / exporting career data would also increase its utility.