dsa-agent: Automated DSA Problem Recommendation and Tracking

Description and use

A developer built dsa-agent to automate DSA problem selection. The agent analyzes solve history in Firestore and recommends the next problem, updating a Notion tracker. The architecture uses two agents for reliability.

Problem

Students practicing Data Structures and Algorithms (DSA) spend excessive time deciding which problem to solve next. Generic AI assistants lack context of solved problems, and static trackers (like spreadsheets) cannot recommend next steps, leading to inefficiency.

Outcome

The dsa-agent successfully automates the selection of the next DSA problem based on personal history, reducing decision fatigue. It integrates with GitHub for automatic tracking and uses a two-agent architecture to ensure reliable output formatting.

Workflow

• Developer commits a solution to the repository via terminal.
• GitHub webhook triggers a FastAPI backend.
• Backend logs the solved problem to Firestore (persistent storage).
• A two-agent ADK pipeline is initiated.
• Agent 1 reads the solve history from Firestore and determines the next recommended problem.
• Agent 2 formats the recommendation into clean output (separated to avoid schema/tool conflicts).
• The recommendation, including a dynamically generated LeetCode link, is displayed on the UI.
• The Notion tracker is automatically updated to mark the problem as solved.

Evidence

When I commit on the terminal, it fires a GitHub webhook that hits a FastAPI backend. The backend logs the solved problem to Firestore...

Source

Splitting it into two agents one focused purely on reasoning over history, the other purely on formatting fixed that.

Source