SkillVault

SkillVault preserves expert employees’ practical decisions and adapts them into traceable, step-by-step guidance for new employees. This helps teams solve issues faster and escalate safely.

Comment

Inspiration

I wanted to make something that solved a real problem that I've seen or heard of others facing. I didn't want it to be super basic, but at the same time did not want to make an overly complex system. After bouncing ideas off Google and ChatGPT, I decided on SkillVault.

What it does

My project takes previously made decisions and work from older, more seasoned employees and saves it in a database for newer, less experienced employees at that company to reference. When people leave companies, all their knowledge, some work, and their overall expertise leave with them. This leaves other workers, both old and new, in need of help with their day-to-day activities.

How we built it

SkillVault was made with Python and Streamlit for a fictional B2B SaaS company called Northstar Cloud. It uses scikit-learn to search/classify past expert decisions, then creates tailored step-by-step guidance. Codex helped with the coding, debugging, data, and design. GPT-5.6 helped with planning and improving the project.

Challenges we ran into

The main issue I had was training the model to adjust to user feedback about how good a job the recommendations from the model were. Also, making a database of knowledge that would then be used to help new employees was hard to implement.

Accomplishments that we're proud of

We created a working product that finds relevant expert decisions, shows its sources, adapts to new problems, allows feedback, and allows employees to add new decisions with optional media. The knowledge pack covers technical support, coding, reporting, presentations, security, billing, and client issues for the fictional company in the demo.

What we learned

That finding a similar decision made before is not enough. The system must explain what changed, provide clear steps, show its sources, and ask for more context when needed.

What's next for SkillVault

Next would be improving the model overall and testing it on current companies in different industries. Each company would have different knowledge stored in the database and different ways of storing that data. Also, implementing security measures to ensure sensitive data doesn't get out and assigning clearance levels to certain data are necessary.

Share this project:

Updates