now buildingSadhak AIstreaming gemini

Softwarethat feelsconsidered.

I'm Rohit Bhatt, a full-stack engineer shipping thoughtful web products with Ruby on Rails, React, and AI integrations.

5+ yearsremote · Indiaopen to work
view work →get in touch
Rohit Bhatt
Uttarakhand, IN
scroll
01 —

Selected work

all projects on github
012025

Sadhak AIai · latest

A privacy-first AI chat that streams Gemini's answers token by token.

Google OAuth, per-user threads persisted in Postgres, and a rolling-summary context window so long conversations stay coherent without blowing the token budget.

Gemini
model
SSE
transport
Postgres
threads
flasklangchaingeminipostgressse
source
022024

Stay a While

A home-stay marketplace — discovery, booking, and host dashboards end to end.

A Rails API feeding a React + Redux client: search and filtering, a multi-step booking flow, host management dashboards, and a clean payments UI.

Rails
api
React
client
Stripe
payments
railsreactreduxpostgres
live demo source
more / open source
space_hub— spacex mission selectorgithub budget_app— rails finance managergithub recipe_app— cookbook + ingredientsgithub finance_app— stock market analysisgithub math_magician— calculator + practicegithub
02 —

About

I build software that feels considered — where the engineering disappears and the experience is all that's left.

Currently working remote, shipping everything from landing pages to full-stack applications, with a recent focus on LLM-backed tools built on LangChain and Gemini.

Five years in, I still care most about the small things: a transition that feels right, an API that's a pleasure to use, a page that loads before you notice. Off-screen, you'll find me on a hiking trail or inside an Erin Morgenstern novel.

download résumé
stack
Ruby on Railsdaily
React · Reduxdaily
Python · Flask
LangChain · Gemini
PostgreSQL
Tailwind · Node
03 —

Writing

read all on /blogs
AnthropicJun 09, 2026

Claude Fable 5, Mythos 5, and the State of the Claude Ecosystem in Mid-2026

On June 9, 2026, Anthropic broke its own naming convention for the first time since Claude 3: Claude Fable 5 is a new "Mythos-class" tier above Opus, generally available at $10/$50 per million tokens, with a sibling — Claude Mythos 5 — reserved for vetted cyberdefenders with some safeguards lifted. This post does two things. First, it explains what Fable 5 and Mythos 5 actually are: the capability claims, the safeguard architecture (high-risk queries silently fall back to Opus 4.8), and what changes for you at the API level. Second, it maps the verified timeline that got us here — Opus 4.5 (Nov 2025), Sonnet 4.6 (Feb 2026), Opus 4.8 (May 2026), Claude Cowork (Jan 2026), Claude in PowerPoint (Feb 2026) — plus the agent stack (Claude Code, Agent SDK, Skills, MCP) that all of it runs on. There's also a meta-lesson: this post started as a research doc in which half of these products were flagged as "probably hallucinated" because they broke known patterns. Every one of them turned out to be real. The takeaway for engineers working with LLMs is at the end.

read
Claude CodeJun 07, 2026

The Definitive Claude Code Playbook (June 2026 Edition)

Most engineers install Claude Code, prompt it like a chatbot, and wonder why it's inconsistent. The answer isn't a better prompt — it's infrastructure. The real gains come from three things: a CLAUDE.md that actually enforces your standards, a fleet of parallel agents in git worktrees instead of one babysit session, and hooks plus sub-agents that handle the tedious parts automatically. The 2025–2026 story is that Claude Code became a platform: 8-event hooks, the renamed Claude Agent SDK, claude-code-action for GitHub, and --dangerously-skip-permissions — great when sandboxed, genuinely dangerous when not. Yes, people have deleted their .git folders with it. Highest-ROI moves for senior engineers, in order: (1) a real CLAUDE.md with DO-NOT rules, (2) a .claude/commands/ stdlib, (3) PostToolUse formatter + Stop test hooks, (4) code-reviewer and security-auditor sub-agents, (5) 3–8 parallel agents on worktrees.

read
RAGMay 11, 2026

Vectorless RAG: You Probably Don't Need a Vector Database

Everyone building AI features in 2026 seems to follow the same recipe: chunk your documents, push them through an embedding model, store vectors in Pinecone or Weaviate, and call it RAG. But the vector database is often the most expensive, fragile part of the stack — and you don't actually need it. Vectorless RAG is a family of retrieval approaches (BM25, PageIndex, knowledge graphs, text-to-SQL, agentic keyword search) that ground LLM answers in real documents without dense embeddings or approximate nearest-neighbour search. This deep dive covers when to skip the vector stack, the five core vectorless approaches with production Rails code, and the honest trade-offs based on the latest 2025–2026 research.

read
$ echo "let's build something"

Let's make it real.