Architectural breakdowns, AI agent systems, Model Context Protocol (MCP) servers, web engineering, and authentic founder notes published from Junagadh, Gujarat, India.
Journal & Field Notes
Sovereign AI in India: Offline 70B + MCP + DPDP 2026
Sovereign AI in India 2026 runs 70B offline + MCP gateway + DPDP audit ledger — from Junagadh, the VPC stack that keeps data inside Gujarat.
Read the story →Strands Agents AWS: Model-Driven SDK for Any Cloud
Strands Agents from AWS is model-driven, any model any cloud, 6.9K stars — from Junagadh, the production SDK that ships with observability built in.
Read the story →Sunday Setup: 5 ADK Agents That Make Monday Easy
5 ADK agents I run every Sunday at 7pm from Junagadh — inbox triager, social queue, outreach, publisher, briefer — wired Parallel+Sequential with approval.
Read the story →Vibe Coding 2.0: Cursor vs Claude vs Codex in Prod
Vibe coding 2.0 with Cursor, Claude Code and Codex CLI — from Junagadh, 98.4% harness vs 1.6% decision logic and brake that saves prod. The ledger that trave...
Read the story →Zero-Trust AI Agents: JWT, OPA & HITL in Production
Zero-trust AI agents in Aug 2026 need JWT scopes, OPA isolation and HITL — from Junagadh, the 40-loop brake plus ledger that passes audits. The ledger that t...
Read the story →A Day in My Life: AI Developer Routine Gujarat 2026
A day in my life as an AI developer in Junagadh, Gujarat 2026: deep work, MCP shipping, factory sprints, and the routine that ships swarms daily.
Read the story →AI Swarms for Indian SMEs: 30-Day ROI Architecture
AI swarms for Indian SMEs that pay for themselves in 30 days: architecture, metering, and Gujarat factory patterns with WhatsApp and GST ROI.
Read the story →Autonomous QA Swarms Cut Bugs 87% in CI/CD 2026
How autonomous QA swarms cut production bugs 87% in 2026: Architect, Security, Test Gen, and Auto-Patch agents in CI/CD with MCP server and 82s gate.
Read the story →Building AI-Native Web Applications in 2026: Architecture, Streaming UX, and Server-Sent Agent Workflows
A complete blueprint for engineering AI-native web apps: Server-Sent Events (SSE), real-time agent state management, vector cache layers, optimistic UI, and...
Read the story →Building AI Products from Junagadh: Real Playbook
Building elite AI products from Junagadh, Gujarat: the Tier-3 city playbook — ROI focus, hiring, infra, and shipping global AI swarms for real clients.
Read the story →Hybrid Reasoning Models: Claude 3.7 & DeepSeek R1
Hybrid reasoning in 2026 explained: thinking budgets, DeepSeek R1 open-weight RL, and how we route tasks between fast and deep reasoning in production.
Read the story →Laravel 13 in 2026: 98 Lighthouse Without a Single SPA
Laravel 13 in 2026: how we hit 98 Lighthouse without an SPA — monolith + Octane, Edge SSR, sub-500ms TTFB, AEO-ready markup for teams under 30.
Read the story →Local LLMs Offline in India: 70B Models on Your Laptop
Run 70B reasoning models offline in India: quantization, GGUF/EXL2, edge SLMs, and the sovereign AI stack we deploy for regulated clients in 2026.
Read the story →MCP Servers in Production: Enterprise Architecture Guide
Production MCP server architecture in 2026: gateway, RBAC, lifecycle, observability, and the enterprise patterns we use in Junagadh for sovereign AI.
Read the story →RAG 2.0 to GraphRAG: What Actually Works in Production
RAG 2.0 vs GraphRAG in 2026: vector, graph, and hybrid retrieval patterns we use to hit <0.1% hallucination for client knowledge bases in production.
Read the story →