Introduction
Week 2 brought clarity to our vision. After conversations with our project advisor Anna, we pivoted from a job-matching platform to something more powerful: an interactive career roadmap tool with AI guidance. This shift addresses the real problem—information fragmentation in skilled trades navigation.
Design Team Updates
Reagan made significant progress on the user survey, incorporating feedback from both Wim (our design teacher) and Anna. She successfully cloned our Git repository and set up her local development environment, preparing to contribute to the codebase. Bruno conducted a crucial feedback meeting with Wim and revised the survey based on Anna's insights before sending it out to potential users.
The team made a strategic decision to focus on one main feature for our December 23rd deadline: a career roadmap for electricians with centralized resources. Darrel continued refining survey questions to ensure we capture the right user insights.
Development Team Updates
Nikita researched RAG implementation extensively, exploring pgvector, PostgreSQL's pg_trgm, and various embedding models including HuggingFace's all-MiniLM-L6-v2 and OpenAI's text-embedding-3-small. Ozem investigated Upstash's vector database and watched educational content on LlamaIndex, while familiarizing himself with AWS services (EC2, S3, Lambda).
Josh had a breakthrough conversation with Claude AI, clarifying Panday's concept: a roadmap visualization tool where users input their goals (e.g., "I want to be an entrepreneur in electrical construction") and receive an interactive, AI-guided path with resources at each node. Peter created a simple LlamaIndex example, demonstrating how we could implement document indexing and retrieval.
TL;DR
We pivoted to a career roadmap visualization tool with AI guidance, moving away from job matching. The team researched RAG implementation, refined our user survey, and defined our core feature: interactive career pathways for electricians.
What's Next?
Week 3 will focus on vector embedding implementation, competitive analysis, and developing our first lo-fi wireframes. We'll also begin analyzing survey results as responses come in.