AI-Pandit: Autonomous BTR Engine

⚠️ UNDER DEVELOPMENT — APIs, architecture, and behavior may change without notice.

⚠️ PROPRIETARY SOFTWARE — ALL RIGHTS RESERVED This repository is publicly visible for transparency and portfolio purposes only. No license is granted to use, copy, modify, or distribute this code. See LICENSE for full terms.

AI-Pandit is a high-performance, autonomous Birth Time Rectification (BTR) platform for the determination of accurate birth times down to the second. It combines classical Vedic astrology with modern LLM reasoning (DeepSeek/Gemini) and NASA JPL DE440 ephemeris data.

---

🏗 Architecture Overview

graph TD
    User((User/Client))
    
    subgraph "Frontend Layer (Vercel)"
        WebApp[Next.js 15 Dashboard]
    end
    
    subgraph "API Layer (Cloud Run)"
        APIService[Express API]
        SSE[SSE Stream Handler]
    end
    
    subgraph "Processing Layer (Cloud Run)"
        Worker[Job Worker]
        BTR_Engine[Vedic BTR Engine]
        AI[DeepSeek Reasoner]
    end
    
    subgraph "Infrastructure"
        DB[(Neon Postgres)]
        Redis[(Upstash Redis Pub/Sub)]
        Skyfield[Skyfield Ephemeris Service]
    end
    
    User <--> WebApp
    WebApp <--> APIService
    APIService -- "Queue Jobs" --> Redis
    Redis -- "Pick Jobs" --> Worker
    Worker -- "Execute" --> BTR_Engine
    BTR_Engine -- "Calc" --> Skyfield
    BTR_Engine -- "Reason" --> AI
    Worker -- "Publish Events" --> Redis
    Redis -- "Sync Bridge" --> SSE
    SSE -- "Real-time Updates" --> User
    APIService <--> DB
    Worker <--> DB

Loading

---

📋 Table of Contents

• Overview
• Features
• Tech Stack
• 6-Stage BTR Pipeline
• Repository Map
• Quick Start
• Environment Variables
• Deployment
• Security
• Testing
• License
• Contact

---

🌟 Overview

AI-Pandit replaces subjective manual BTR with a data-driven autonomous pipeline. It generates thousands of candidate birth times, runs them through successive AI-supervised elimination rounds (Dasha verification, transit matching, KP Sublord analysis, Shadbala evaluation), and converges on the most astronomically and astrologically consistent time.

The system processes life events as input constraints, cross-references against JPL DE440 ephemeris via Skyfield, and encrypts all PII with AES-256-GCM end-to-end.

---

🚀 Features

• Autonomous 6-Stage Pipeline: Grid Generation → Batch Tournament → Refinement → Deep Analysis → Micro Grid → Final Verdict.
• Real-Time SSE Streaming: Live progress updates with Zustand + IndexedDB persistence.
• NASA-Precision Ephemeris: Skyfield service providing arcsecond-precision planetary positions.
• AI-Driven Reasoning: DeepSeek-Reasoner model for complex astrological synthesis.
• End-to-End Encryption: AES-256-GCM with user-isolated keys and multi-version format support.
• Interactive Dashboard: Full session management, PDF exports, and Recharts visualizations.

---

🛠 Tech Stack

Layer	Technology
Frontend	Next.js 15, React 18, Zustand, Tailwind CSS, Framer Motion, Recharts
Backend	Node.js (Express), TypeScript, Drizzle ORM, Zod
Database	Neon Postgres (Serverless)
Cache/Queue	Upstash Redis (ioredis)
AI	Groq (GPT-OSS-120B), DeepSeek (Fallback)
Auth	Clerk (OAuth/MFA)
Ephemeris	Python 3, FastAPI, Skyfield (DE440 Kernel)
Deployment	Vercel (Web), Google Cloud Run (API/Worker/Ephemeris)

---

⚖️ 6-Stage BTR Pipeline

Stage	Name	Description
1	Grid Generation	Generate exhaustive candidate time grid around tentative birth time
2	Batch Tournament	AI-supervised batch elimination — prune clearly incompatible candidates
3	Refinement Grid	Sub-second finer grid around remaining survivors
4	Deep Analysis	Multi-dasha, multi-transit cross-validation with life events
5	Micro Grid	Seconds-level grid with precision ephemeris data
6	Final Precision	AI synthesis of all evidence → final rectified time + verdict

---

🗺 Repository Map

ai-pandit/
├── apps/
│   ├── web/                      # Next.js 15 frontend dashboard
│   ├── api/                      # Express + TypeScript BTR orchestrator
│   └── worker/                   # External background job worker
├── packages/
│   ├── db/                       # Drizzle schema + client (Neon)
│   ├── shared/                   # Shared Zod schemas and TS types
│   └── worker-runtime/           # Shared worker processing library
├── services/
│   └── ephemeris/                # Python FastAPI Skyfield microservice
├── e2e/                          # Playwright end-to-end tests
├── scripts/                      # Deployment and utility scripts
├── .github/                      # CI/CD workflows and templates
└── AGENTS.md                     # Agent operating manual

---

⚡ Quick Start

1. Installation

npm ci
npm run setup:ephemeris
npm run ephemeris:download-kernel

2. Environment Setup

cp .env.example .env.local
# Edit .env.local with your keys (Clerk, Neon, Redis, DeepSeek)

3. Database & Dev

npm -w @ai-pandit/api run db:push
npm run dev

---

🔒 Security

• PII Encryption: AES-256-GCM at rest/transit. Key derived via scrypt with user-specific salt.
• AI Anonymization: All prompts are stripped of names and exact birth locations before inference.
• Auth Hardening: Clerk-managed session tokens, MFA support, and CSRF protection.
• Infrastructure: All services deployed with identity-aware IAM on Google Cloud.

---

🧪 Testing

npm run test           # All unit tests
npm run test:integration # API + DB integration
npm run test:e2e:smoke  # Critical path E2E
npm run test:security   # Dependency and secret scan

---

📄 License

Proprietary. See LICENSE for full terms. No license is granted for use, modification, or distribution.

---

📬 Contact

Author: Ashok Saini
Email: app.aipandit [at] gmail [dot] com
Repository: github.com/ashoksainiengineer/ai-pandit-app

---

Built with ❤️ for the Vedic astrology community