Skip to content
View DhanushAleti's full-sized avatar
  • India

Block or report DhanushAleti

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
DhanushAleti/README.md
Dhanush Aleti — AI Orchestrator, founder of Raha



Raha — live LinkedIn Email GitHub



profile views

About

I'm a founder and an AI orchestrator. I build companies the way you'd build a system: models, tools, and agents do the execution, and my job is deciding how the pieces fit, where the handoffs happen, and what to do when one of them fails.

I run a one-person company. That means I ship the product, write the code, do the research, and talk to the customers — with an AI stack doing the volume work underneath. Right now all of that points at Raha.

I build in public. Most of what I learn ends up as a repo here, including the rough parts.


Raha — creator taxes, handled

raha-iota.vercel.app · live in production

Tax and GST compliance built for Indian creators earning ₹20L–₹2Cr — the band that has outgrown a spreadsheet but isn't big enough for a full-time finance team.

  • FIRC tracking for foreign-income receipts
  • GST-compliant invoicing built to the actual filing format
  • A live "set aside this much" counter, so tax stops being a year-end surprise

Every figure is presented as an estimate to be confirmed with a qualified CA — the product assists with compliance, it does not file on your behalf.

Built solo, end to end: product definition, data model, auth, billing surface, and deployment. Next.js · TypeScript · Supabase · Vercel · 68 unit tests, production build green.

Source is private — happy to walk through the code or architecture.


Selected Work

Project What it is Stack
Exoplanet Hunter (TESS) Transit-search pipeline over real NASA TESS light curves. Queries MAST, stitches and flattens light curves, runs a BLS periodogram, and logs candidate signals to a catalog. Python Lightkurve NumPy
GPT From Scratch A character-level GPT built from the ground up — tokenizer, bigram baseline, self-attention, and a small transformer. Implemented step by step rather than via a high-level library. Python PyTorch
Code Debugger Agent An AI agent that analyses code, diagnoses failures, and proposes fixes — built to shorten the debug loop. Python LLM tooling
RAG Pipeline Retrieval-augmented generation over a document corpus: chunking, embeddings, vector retrieval, and grounded answer synthesis. Python Embeddings
FounderFlow AI Toolkit · live ↗ The prompt and system layer behind how I run a one-person company — structured tools for thinking, validating, and growing. Prompt systems
EdList Dashboard · live ↗ Project-management dashboard with a modern App Router frontend, typed end to end and deployed on Vercel. Next.js 15 TypeScript Tailwind
OrbitalMind Orbital-mechanics and astrodynamics workbench — trajectory reasoning and simulation-style analysis. Python SciPy

Also In Progress

Private while they take shape.

Project What it is
Project PHI AI-first product, full-stack, in active development.
Founder Copilot An AI second brain for founders — search, chat, and reason over startup documents with RAG.

Stack

languages and frontend
backend, data and infra

Activity

contribution streak



contribution activity graph
contribution snake

Get In Touch

Talk to me if you're an Indian creator dealing with GST and foreign income — that's who Raha is built for, and I want to hear where it falls short. Also open to design partners, collaborators, and anyone building at the intersection of AI and real products.


Building in public from Hyderabad, India.

Pinned Loading

  1. gpt-from-scratch-dhanush gpt-from-scratch-dhanush Public

    Built a GPT-style language model from scratch in PyTorch by following Andrej Karpathy’s tutorial, including tokenization, bigram baseline, self-attention, and a small transformer.

    Python