GoSkill

InspirationAbout the project

Inspiration

Choosing a new career path can be overwhelming — and figuring out the skill gaps between where you are and where you want to be isn't always obvious. I wanted to make that easier with AI. Specifically, this was inspired by all the posts online about it being so challenging to get a new job.

What it does

GoSkill lets you upload your resume or describe your current skills, describe your dream job(location working hours, you name it), and instantly receive:

• A recommended best-match role
• Detailed info like location, compensation, responsabilities, etc.
• Your personalized skill gap analysis (ranked and categorized)
• Suggested Coursera courses to close the top gaps
• A growth path roadmap toward leadership or specialization

How I built it

• Streamlit powers the interactive web app through localhost
• Hugging Face Transformers (zero-shot classification) categorizes skill gaps
• Sentence Transformers (MiniLM) and TF-IDF provide semantic search and similarity scoring
• Pandas and Scikit-Learn help process and match job descriptions
• Coursera xAPI fetches course recommendations for missing skills
• PDF and DOCX parsing allows resume uploads
• Session state and caching make the app fast and responsive

Challenges I ran into

• Building a recommendation engine that worked both pre-trained using in-depth labels
• Parsing resumes reliably across PDFs, DOCX, and text files
• Categorizing free-form skills into business/technical/certifications using only minimal API calls
• Handling API limitations like rates

What I learned

• How to combine semantic search (SBERT) with classic TF-IDF in a fallback-friendly way
• How to cache large model loads and precompute embeddings for a fast UX
• Designing multi-step Streamlit apps with decent UX
• Using zero-shot classification methods to find the relevance and thus the importance of a skill.
• Scraping skills from Job data and classifying them.

What's next

• Adding LinkedIn profile related things
• Expanding course recommendations to other providers like edX or Udemy
• Building "mini skill plans" based on user timelines
• Use real-time job-search APIs

Built With

• all-minilm-l6-v2)
• api
• coursera
• docx2txt
• face
• facebook/bart-large-mnli)
• hugging
• pandas
• pdfplumber
• python
• scikit-learn
• sentence
• spacy
• streamlit
• transformers