Hamin Hesab
An AI-assisted accounting platform that automates the boring 80% of personal and small-business finance — from a Telegram chat to a ledger.
Senior Software Engineer & PM
Timeline
2024
Team
Senior engineer + PM (myself) + product
Role
Senior Software Engineer & PM
Overview
Hamin Hesab lets individuals and small teams keep clean books without opening a spreadsheet. A Telegram-first interface, natural-language transaction capture and an intelligent decision-support layer replace most of the manual reconciliation work.
The problem
Small businesses in Iran either skip bookkeeping entirely or pay for tools built for accountants, not owners. The gap between how people actually talk about money and how ledgers demand it costs hours every week.
Approach
I led backend architecture, the Telegram bot integration, and the decision-support layer that turns free-form messages into typed accounting entries. The system uses an LLM front-end for parsing plus deterministic post-processing so numbers never depend on a hallucination.
— What I built
Key capabilities.
- 01Telegram-first capture — text, voice notes, and receipt images
- 02LLM parser that maps natural-language messages to typed transactions
- 03Automatic categorisation with per-user learning from corrections
- 04Reports and cashflow views generated from the ledger, not the model
- 05Multi-organisation support with role-scoped access
— Outcomes
What shipped.
Primary surface
Telegram
no app install
Entry types
12+
typed transaction schemas
Safety model
LLM in / rules out
no hallucinated numbers
— This project
Live and in production.
Open the live project or start a conversation about collaboration.
— More work
Digital Platform
Bikaran
A digital services platform I designed, built, deployed and operate alone — from product strategy to VPS — now serving 700–1200 monthly active users.
Read case study→
LLM · Legal Assistant
TRSKHOOB
An AI-native legal assistant that turns messy natural-language questions into precise, source-grounded consultation flows — in Persian.
Read case study→