Behavioral Analytics Bot with Exchange API Integration
A Telegram bot that connects to a financial platform over a read-only API, pulls transaction history, computes behavioral metrics, and delivers weekly reports generated by an LLM. Subscription billing built in.
- Role
- Backend Engineer
- Evidence
- Read-only integration · weekly LLM reports · subscription billing
- Status
- Private by design
The problem
People making financial decisions rarely see their own patterns. The data exists, but nobody reads their own history.
What I built
A service that reads the data for them and reports what it finds.
- Read-only API integration. Pulls transaction history without any ability to move funds, which removes the largest trust barrier for a product like this.
- Normalisation layer. Records arrive from different sources in different shapes and are mapped into a single internal model before any metric is computed.
- Rolling metrics. Behavioral indicators calculated across time windows, with scheduled background workers producing weekly summaries.
- LLM reporting. Computed metrics are turned into readable narrative reports rather than a wall of numbers.
- Subscription billing. Payment flow implemented through Telegram Stars.
Engineering standards
Strict typing with static analysis enforced in the pipeline. The service handles financial data and API credentials, so type errors are caught before deployment rather than in production.
Stack
Python, aiogram, PostgreSQL, Redis, LLM API, Docker.
Proprietary product, source closed.
Ready when the problem is real
Where does manual work become expensive for you?
Describe the process, the failure you cannot afford and the result you need. We will propose the smallest system that solves it reliably.