Inspiration
Searching through hundreds of job listings across many different sites to find which ones are relevant for your interests and a good match for your skills is incredibly tedious and time consuming. We wanted to utilise the power of LLMS to create a much more streamlined, automated and intelligent way of conducting this process. We also wanted to leverage its abilities to help people keep up to date and upskill in areas of interest to them, to further help with their career progression.
What it does
By automatically parsing a resumé submitted in pdf form, C3 uses Claude to build a profile of the user's professional working experience and skillset. It then scrapes popular job listing sites, filtering by keywords determined by the profile for relevant jobs. Going a step further than sites like Indeed or LinkedIn, C3 will then analyse the text of those job descriptions and compare it against the user's profile to figure out if there is a good match between the user's skills and experience and that expected for the role. It then utilises Claude again to produce a short synopsis of why it thought the user would be a good for for the role, and what unique features the user has that might make them distinguished for the role, useful tools for potential interviews and questions.
In parallel, it compiles the user's key skills and skills it thinks would be relevant to learn and searches the internet for relevant and recent articles in this area, producing a newsletter which is emailed weekly.
How we built it
Using Anthropic Claude 2 LLM for textual summarisation and analysis tasks, REACT for the front end and Firebase for the serverless back end.
Challenges we ran into
Web scraping proved much more difficult than expected, as a number of these sites require sign up or have anti-scraping tech. Connecting the front end and back end within the time limit was also very tough. Also, as the time for a Claude completion scales with its prompt size, the time taken to fulfil some of the large requests was quite long.
Accomplishments that we're proud of
Learned how to use a number of unfamiliar APIs in a very short time.
What we learned
Realised the value of tools such as Langchain that simplify a process we built manually due to inexperience.
What's next for C3 - Claude Career Compass
Complete the newsletter function and connect it all up. Expand into other areas for career progression assistant - virtual CV / cover letter reviews, mock interviews etc.
Log in or sign up for Devpost to join the conversation.