❯AI Engineering Assistant — Nine Phases
Multi-phase LLM agent workflow: a nine-stage pipeline that plans, retrieves, reasons, acts through tools, and verifies before it ships an answer.
- Python
- LangGraph
- RAG
- tool-calling
source ↗Python
open to senior backend / AI roles · tehran, ir
Sina Soleymanzadeh
senior backend & ai systems engineer
Six-plus years designing and scaling distributed backends — Node.js and Go services, real-time systems at 10,000+ concurrent sessions — now shipping production LLM systems: agent workflows, RAG pipelines, and the infrastructure between them.
sina@tehran — zsh
sina@tehran:~$ whoami
senior backend & AI systems engineer — 6+ yrs
sina@tehran:~$ ls ~/stack
node/ go/ typescript/ python/ nestjs/ langgraph/ postgres/ redis/ k8s/
sina@tehran:~$
~$ cat about.md
About
I'm a backend engineer with 6+ years designing, building, and scalingdistributed systems — Node.js and Go services for financial applications, real-time platforms holding 10,000+ concurrent sessions, and the data layer underneath: PostgreSQL, MongoDB, Redis, Elasticsearch.
Lately I've been deep in production AI systems — multi-phase LLM agent workflows and RAG pipelines that answer with sources instead of vibes. That work lives in the open on GitHub: agent pipelines, AI-driven commerce, censorship-circumvention tooling.
I've led monolith-to-microservices migrations, built CI/CD pipelines from zero, and run production on Kubernetes. I live in the terminal, self-host what I can, and am based in Tehran, Iran.
about.toml-rw-r--r--
[engineer]
location = "tehran.ir"
experience = "6+ years"
os = "arch linux" # btw
editor = "doom-emacs"
shell = "zsh + tmux"
[defaults]
apis = "node + go"
agents = "python"
hosting = "self"
uptime = "boring" # the good kind~$ tree ~/skills -L 1
Skills
skills/languages
# daily drivers first- TypeScript / JavaScript
- Go
- Python
- SQL
- C / C++
- Rust (learning)
- Bash
skills/backend
# services that stay up- Node.js / Express
- NestJS
- Echo (Go)
- REST & GraphQL APIs
- WebSockets / Socket.IO
- microservices
- clean architecture
- real-time systems
skills/ai-llm
# production, not demos- LangGraph
- LangChain
- RAG pipelines
- agentic tool-calling
- embeddings & vector search
- prompt & eval workflows
skills/databases
# boring and reliable- PostgreSQL
- MongoDB
- Redis
- Elasticsearch
- Qdrant (vector)
- TypeORM / Sequelize / GORM
skills/devops-tooling
# self-hosted first- Docker / Compose
- Kubernetes
- CI/CD pipelines
- Nginx
- Grafana observability
- Cloudflare Pages / Workers
- Linux (Arch, btw)
- Doom Emacs + tmux
$ man sina | grep -i "currently learning"
→ rust — because the borrow checker builds character
~$ ls ~/projects --sort=impact
Featured Projects
❯sale-ai
AI-driven sales and commerce system — LLM agents wired into product data, pricing, and order flows to answer like a rep who actually read the catalog.
- Python
- LLM agents
- commerce
source ↗jul 2026Python
❯v2ray-ip-replacer
Censorship-circumvention tooling for V2Ray users — finds fresh working IPs and swaps them into client configs, keeping connections alive on hostile networks.
- V2Ray
- networking
- censorship circumvention
source ↗may 2026HTML
❯guess-my-draw
Real-time multiplayer drawing-and-guessing game — WebSocket state sync, rooms, and low-latency canvas broadcast between players.
- JavaScript
- WebSockets
- realtime
source ↗JavaScript
❯nestjs-blog-backend
Blog platform backend in NestJS — modular architecture, auth, and a clean REST surface. The framework I run in production, in the open.
- TypeScript
- NestJS
- REST
source ↗TypeScript
❯tolling
Distributed toll-calculation system in Go — small event-driven microservices wired together, an exercise in service boundaries and data flow.
- Go
- microservices
- event-driven
source ↗Go
❯clog
Weblog backend API in Express + TypeScript — routing, auth, and persistence with clean separation between layers.
- TypeScript
- Express
- REST
source ↗TypeScript
$ gh repo list cna-so — all repositories ↗
# stars & metadata pulled from the GitHub API at build time
~$ git log --oneline --graph career
Experience
f4b8e2a (HEAD -> now)Feb 2025 — now
Backend Developer — Bitium Agency (Dubai)
Backend microservices for critical financial applications in Node.js and Express — built for high availability and scale. Cut data-retrieval latency by 25% through MongoDB and PostgreSQL query optimization, and built the full CI/CD pipeline that tests and ships all of it.
- Node.js
- Express
- PostgreSQL
- MongoDB
- CI/CD
9c1d4e0Feb 2024 — Feb 2025
Backend Developer — AkafWeb (Istanbul)
Real-time backend for a major Finnish ride-sharing service: NestJS + Socket.io handling 10,000+ concurrent sessions at peak. Built the monitoring stack on Grafana and Elasticsearch, and cut average API response time by 35% with Redis caching strategies.
- NestJS
- Socket.io
- Redis
- Grafana
- Elasticsearch
7a3f9b1Dec 2022 — Feb 2024
Front-end Developer — Moniaz (Tehran)
Led a re-architecture to Next.js and TypeScript that lifted LCP by 40%, refactored state management to halve redundant API requests, and replaced a third-party video player with a custom high-performance one. Backend empathy, earned from the other side of the API.
- Next.js
- React
- TypeScript
- performance
2e8c6d5Jul 2020 — Dec 2022
Full-Stack Developer — Targoman Intelligent Processing (Tehran)
Re-architected a monolith into a scalable, maintainable system with modern design patterns. Containerized everything with Docker, orchestrated deployments on Kubernetes, and optimized PostgreSQL schemas and data-processing paths. Where the systems habit started.
- Docker
- Kubernetes
- PostgreSQL
- architecture
~$ ping cna-so
Contact
PING sina.tehran.ir: 56 data bytes
64 bytes from tehran: icmp_seq=0 ttl=64 time=fast
64 bytes from tehran: icmp_seq=1 ttl=64 time=<24h # email response time
--- sina ping statistics ---
2 packets transmitted, 2 received, 0.0% packet lossBuilding something with LLMs that needs a backend that doesn't fall over — or just want to talk Go, agents, or proxies? My inbox is open.
Tehran, Iran · IRST · based in Tehran
open to senior backend / AI systems roles — remote or on-site in Tehran