NLP Libraries for Resume Optimization?
NLP Libraries for Resume Optimization?
Hey everyone,
Recently, I've found myself bogged down in sending off resumes that seem to never to be read by anyone other than myself.
I'll go through the whole gamut of picking keywords that match the job description, showcasing my previous experiences, projects, skills etc... But it just seems to never result in a call-back or even an email to tell me I wasn't selected.
Given that I'm tired of screaming into the hills and hearing it echo back, I want to write a program that streamlines this whole process. I have a couple of resume templates written in TeX script that I can populate with content. Alongside this, I have all of my relevant bullet points in assorted text files labeled appropriately.
The idea would be to feed the program the job description, relevant qualifications, and other miscellaneous text files. These would be processed to give an idea on how my resume should be modified to suit their requirements. Perhaps that could aid in creating a strong resume in a more streamlined fashion. I have no clue what metric should be used to quantify how "good" it is, so that's to be figured out as well.
I saw "nltk" and "spaCy" are two NLP libraries for Python, but I wanted to open up discussion for those of you who have worked on projects similar to this. I have read mixed comments about the two. Which one seems better suited for this task?
Obviously I'll review the resume before I submit it, but I want to see if I could get something like this working.
I'm a giant noob when it comes to NLP, but have used Python for the past couple of years for data-science applications. I'd be open to learning a different language if there is a library that has some of these functions already coded, but I'm not a developer.
Thanks for any help! I love the community over here on Lemmy. Many of you have been very helpful and encouraging and it makes me want to keep learning more :)
Put this on GitHub and include it in your resume.
Maybe subtly with a watermark in 5pt font that says "Made with Resume-Strapper" lol
If I'm understanding you correctly, you might want to take a look at some NERT models for topic extraction on huggingface. I'm not certain there are already ones for resume building, but I'm sure they exist. An alternative is spaCy, as you mentioned, but there you also rely on a pretrained model.
It really sounds like you’re looking for an LLM text generator. I prefer using direct NLP for many NLP related tasks, but I think what you’re describing is directly related to what LLMs do.
Of course, neither the NLP approach nor the LLM ones will benefit from a feedback loop of successful versus unsuccessful resumes. Advertisers and spammers can generate a thousand variations of the same message and throw them all out onto the internet, then just reinforce the successful ones and drop the others. That works because there’s a single ad shop and hundreds of millions of users who don’t get together to compare their Temu ads. You, on the other hand, are dealing with a handful of employers, and each of them may have different criteria. A resume that gets you a job coding at a bank is t the same as one that will get you in the door at Apple or Google. You’ll have to reinforce with mostly gut feeling or the help of friends.
Unfortunately I don't think completely automating my resume is going to happen. It's just a dream :( I've finally found something that got the attention of an employer though, so hopefully my job search will be over soon.
I'm still itching to do something with NLP/LLMs, but I'll have to define the problem more rigorously rather than throw out nebulous desires. Thanks for the response!