~/ilir-hushi
full-stack engineer

Ilir Hushi

Available for new projects

I build the backend systems behind payments, crypto accounting, and AI products, and keep them observable in production. Node.js and Python, from Kafka topologies to the React Native apps running on top.

Get in touchSee recent work
01 · about

Background

I've been writing software for over nine years, mostly on the backend. Node.js and Python are my primary tools. I've built a lot of distributed systems along the way: event-driven architectures, microservices, the observability layer that tells you when things go wrong. I'm just as comfortable designing a Kafka topology or tuning a pgvector index as I am shipping a React Native app to iOS and Android. Lately I've been building AI-powered backends too: MCP servers, RAG pipelines, semantic search on Kubernetes. I've always worked remotely. Outside of code, you'll find me playing or watching football. Lifelong AC Milan supporter. Once a Rossonero, always a Rossonero. Forza Milan.

02 · stack

Running services

backend

  • Node.js
  • Python
  • FastAPI
  • PostgreSQL
  • Redis

infra & data

  • Kafka
  • Docker
  • AWS
  • GCP
  • Elasticsearch
  • Kubernetes

frontend

  • TypeScript
  • React
  • React Native
  • Angular

ai & web3

  • pgvector
  • RAG
  • MCP
  • Web3
03 · log

Recent work

SHIPDealflow ↗

Event-driven payment microservices

At Dealflow I built Python microservices with FastAPI and asyncio, integrating Kafka for async event streaming across payment and invoicing flows at scale. The whole thing was instrumented with OpenTelemetry and Grafana for production observability.

high-throughput payment flows in production, fully traced end to end
BUILDGilded Finance ↗

Crypto invoicing & bookkeeping backend

Built REST APIs and event-driven data pipelines for Gilded Finance, a blockchain invoicing platform connecting QuickBooks, Xero, and on-chain wallet data via Web3 and Node.js.

real-time wallet ingestion + two-way QuickBooks & Xero sync, live
SHIPBuletin ↗

AI-powered news platform

Built the MCP server powering Buletin's news analysis: semantic search and LLM-generated summaries using pgvector, RAG, and Ollama embeddings. Deployed on Kubernetes with Prometheus and Langfuse for observability.

semantic search & AI summaries serving readers daily on Kubernetes
BUILDLandmark Technologies ↗

Full-text search & cross-platform apps

At Landmark Technologies I built a custom Elasticsearch search engine with filters and a suggestion engine, and shipped several cross-platform mobile apps in React Native to iOS and Android.

search engine + multiple apps shipped to the iOS and Android stores
04 · contact

Get in touch

GET /contact

Open to new projects and collaborations. The fastest way to reach me is email. I'll get back to you within a day or two.

book a callgithublinkedin