An optimized, production-ready genetic health interpretation engine that scans raw DNA exports (like 23andMe) and generates interactive, bilingual (English + Hindi) health insight reports in real-time. Built specifically to handle large-scale genomic data under serverless hosting constraints.

---

🛠️ How It Works (System Architecture)

Processing raw genome files (~20MB, 600,000+ lines of text) on free serverless platforms (like Vercel Hobby) presents two key engineering challenges:

1. Request Timeouts: Serverless functions have a strict 10-second execution limit.
2. Payload Limits: Maximum POST request payload size is 4.5MB.

To circumvent these limits, this application employs a hybrid browser-edge processing architecture:

[ User Browser ]                                    [ Vercel Edge Serverless ]
┌──────────────────────────────┐                   ┌──────────────────────────────┐
│ 1. Local File Upload         │                   │                              │
│ 2. Chunked Stream Parsing    │   ~1KB Genotypes  │                              │
│    (FileReader API, <100ms)   │ ────────────────▶ │                              │
│ 3. Filter against Local JSON │    (POST payload) │                              │
│    ClinVar SNP Index (27)    │                   │                              │
│                              │                   │ 4. Groq Llama-3.3-70B        │
│                              │                   │    Streaming Analysis        │
│ 5. Streams & Renders         │ ◀─── text/event ─ │                              │
│    Markdown Progressive UI   │        stream     │                              │
└──────────────────────────────┘                   └──────────────────────────────┘

1. Client-Side Stream Parsing: The raw .txt file never leaves the user's browser. The application streams the file locally using the browser's FileReader API, parsing it line-by-line in under 100 milliseconds.
2. Index Filtering: The parsed SNPs are cross-referenced with a local catalog of 27 high-impact, clinically relevant markers (Methylation, Lipid metabolism, Cardiovascular, Inflammation).
3. Payload Reduction: The frontend sends a clean, filtered JSON payload containing only matching risk genotypes (~1KB) instead of the raw 20MB file.
4. Edge Streaming Completion: The edge function streams response tokens from the Groq API back to the client as they generate, bypassing serverless idle timeouts.

---

✨ Features

• 🔒 Absolute Privacy: 100% client-side file parsing. Zero genomic data persists on any server.
• ⚡ Sub-Second Processing: Near-instant client-side parsing of half a million database rows.
• 🤖 Bilingual Agentic Reports: Real-time generation of custom wellness and lifestyle guides in English and Hindi (Devanagari script), using the ultra-fast llama-3.3-70b-versatile model.
• 🚦 Priority Risk Ranking: Matches are automatically categorized and sorted by clinical severity (🔴 High, 🟡 Moderate, 🟢 Low) so action items are addressed sequentially.
• 📑 Rich UI Layout: Elegant glassmorphic dashboard built using TailwindCSS, React, Lucide Icons, and React-Markdown for rendering interactive reports.