DevPulse

Overview

A developer telemetry dashboard tracking cognitive load and productivity trends. DevPulse fires an AI inference call to analyze commit narratives, calculating a developer's sentiment score, burnout index, and "vibe", wrapped in an industrial brutalist aesthetic.

AI Pipeline Strategy

• Groq Inference: Ultra-fast analysis on commit batch data using llama-3.1-8b.
• Burnout Indexing: Formula estimating intensity vs recovery per commit.
• Executive Directive: LLM compiles aggregated stats into a single industrial command summary.
• Rate-limit Backoff: Exponential retries built over Octokit and Groq integrations.

System Architecture & Telemetry Flow

A robust full-stack architecture powered by Express 5 orchestrating asynchronous Git integrations. It scales JSON metrics via authenticated GitHub Passport.js OAuth sessions and securely dumps persistent historic telemetry profiles directly into a Mongo cluster, visualizing seamlessly via React DOM.

sequenceDiagram
    participant UI as React UI 🖥️
    participant Server as Express Backend 💻
    participant Github as GitHub API 🐱
    participant AI as Groq (Llama 3) 🧠
    participant DB as MongoDB Atlas 🗄️

    UI->>Server: Request Telemetry (Token)
    Server->>Github: OAuth Session / Fetch Commits
    Github-->>Server: Return Commit History Batch
    Server->>AI: Send Commits for Inference (System Prompt)
    AI-->>Server: Returns Vibe/Burnout Multipliers
    Server->>DB: Store Analyzed Telemetry Profile
    Server-->>UI: Serve Aggregated JSON Payload
    UI->>UI: Render Recharts Data Visualization
                        

Tech Highlights

Core Engineering Challenges

• AI Strict Formatting: Constrained Llama-3 outputs strictly into deterministic JSON patterns, resolving hallucinated object structures via hard-coded parsing strategies.
• API Backoffs: Implemented smart exponential retries to circumvent strict free-tier rate limits from third-party OAuth callbacks.