MentorUp

Your guide to your next billion dollar venture

Comment

Inspiration

Mentorship in the startup world is often limited to those with the right networks, geography, or financial means. I realized that by aggregating high-quality insights and advice from leading mentors around the globe and organizing them through robust vector indexing techniques, we could democratize access to that knowledge. Bringing together disparate pieces of expert guidance into a cohesive, searchable knowledge base inspired the creation of MentorUp.

What it does

MentorUp enables founders to get instant, personalized answers to their startup questions by leveraging MongoDB Vector Search. When you submit a query, MentorUp:

  1. Generates an embedding of your question using Google Vertex AI.
  2. Searches a curated knowledge base of mentor insights stored in MongoDB with vector search.
  3. Retrieves the exact response to your question from the best mentors.

How we built it

  • Embedding Generation: Integrated Google Vertex AI to produce embeddings for both user questions and mentor content.
  • Vector Search: Utilized MongoDB’s vector search to efficiently match queries against our knowledge base.
  • Knowledge Base: Curated a comprehensive dataset of mentor insights, case studies, and session transcripts.
  • API & Frontend: Built with Next.js and TypeScript for a seamless, responsive web experience.

Challenges we ran into

  • Deployment Complexity:This was my first time deploying on Cloud Run. Setting up reliable deployment pipelines using Docker, and CI/CD workflows, and managing environment-specific configurations across staging and production added new challenges.
  • Indexing Scale: Tuning MongoDB vector search parameters to maintain low latency under load.
  • User Experience: Designing an intuitive interface to guide founders through question formulation and deliver clear advice.

Accomplishments that we're proud of

  • Successfully deployed on Cloud Run with robust CI/CD pipelines.
  • Built an efficient MongoDB vector search index for rapid, relevant retrieval.
  • Created high-quality embeddings through Google Vertex AI for precise semantic matching.

What we learned

  • High-quality data curation is as important as AI model choice—organized content drives better results.
  • Fine-tuning vector search configurations significantly impacts response speed and accuracy.
  • Learned a lot about Google Cloud services and best practices for managing cloud infrastructure.

What’s next for MentorUp

  • Develop personalized learning paths that adapt to each founder’s progress and needs.
  • Add community-driven features like upvoting and commenting to foster peer engagement and a social knowledge graph.
  • Add more high quality data for better response to the user query

Built With

Share this project:

Updates