Backend developer · Tehran, Iran

Pouya
Sadeghi

I design backend systems that stay correct under concurrency—from marketplace synchronization and real-time gold trading to reusable, typed Python infrastructure.

23Public repositories
13GitHub followers
17Following
2021Building in public since
Profile, rendered as code

The engineer behind the systems.

A terminal-style snapshot of how I work: performance-aware, detail-oriented and comfortable moving between architecture and implementation.

00
Matrix terminal showing Pouya Sadeghi's profile with an ASCII butterfly
source: github/profile/identity.svgsignal: verified
Selected production work · Goldis

Backend systems for live commerce.

Three production platforms covering marketplace operations, automated trading and dynamic pricing—designed around concurrency, failure recovery and financial correctness.

01
system_01 · Commerce infrastructure

Multi-Platform Sync & Pricing Engine

production
  • Synchronizes products, inventory, orders, discounts and invoices across Digikala, Snapp, Basalam and Tapsi.
  • Coordinates roughly 12 concurrent workers to process and propagate marketplace data with low delay.
  • Supports Automatic, Competitive and Takeover pricing with live competitor crawling and hard stop-loss limits.
  • Matches external marketplace listings to internal inventory through product mapping and recommendations.
  • Includes a lightweight custom ORM path for database-intensive operations where standard abstractions add measurable overhead.
  • Provides financial endpoints for profit calculation, settlement and operational reporting.
PythonLitestarTaskiq PostgreSQLRedis
system_02 · Fintech

Digital Gold Trading Platform

  • Executes back-to-back supplier purchases immediately after user trades to reduce exposure to gold-price movement.
  • Uses RabbitMQ and Redis workers for reliable, concurrent and asynchronous trade execution.
  • Detects supplier failures and activates operational fallbacks, monitoring and immediate administrator alerts.
FastAPIRabbitMQ RedisPostgreSQL
system_03 · E-commerce

Gold & Jewelry Commerce

  • Reprices store inventory automatically from live gold-market rates.
  • Categorizes products through heuristic scoring while preserving explicit administrative control.
  • Collects and scores training data for a future neural-network categorization model.
FastAPICelery RedisAlgorithms
Open-source engineering

Infrastructure made reusable.

Frameworks and typed SDKs extracted from recurring backend problems: application structure, provider fragmentation, unreliable payloads and sync/async duplication.

02
Code, experiments and long-term practice

Public work, in the open.

The full public footprint—from production-minded Python packages to algorithms, data structures, notebooks and low-level experiments.

03

Backend developer at Goldis · Tehran, Iran

23 repositories 13 followers 17 following Member since 2021
Open GitHub profile ↗
Contribution calendar · last 12 months365D
Pouya Sadeghi's GitHub contribution calendar for the last twelve months
Contribution activity · weekly totalsLINE GRAPH
Line graph of Pouya Sadeghi's weekly GitHub contribution activity
Contribution snake · animatedLIVE
Animated snake moving across Pouya Sadeghi's GitHub contributions
Tools selected by constraints

A practical backend toolkit.

I choose technologies for the workload and failure model—not for novelty. The common thread is typed boundaries, observable behavior and maintainable concurrency.

04
module_01

Backend architecture

Modular services, clean dependency boundaries, CQRS read models and domain-led APIs.

FastAPI · Litestar · Django · Dishka
module_02

Concurrency & messaging

Worker fleets and asynchronous pipelines with explicit retry, failure and throughput behavior.

Taskiq · Celery · RabbitMQ · Redis
module_03

Data systems

Transactional persistence, caching and search projections designed around access patterns.

PostgreSQL · Elasticsearch · SQLAlchemy
module_04

Typed SDK design

Predictable clients that normalize third-party APIs without discarding provider-specific detail.

Pydantic · dataclasses · httpx · aiohttp
module_05

Algorithms

Data structures, heuristics and correctness reasoning carried from contests into production.

C++ · Python · Problem solving
module_06

Low-level exploration

Systems and graphics work that keeps the layers below the framework understandable.

C · Rust · SDL2 · Linux
Start a conversation

Have a backend problem worth solving?

I’m interested in systems where concurrency, correctness and performance are product requirements—not afterthoughts.