Mohammad Hossein Shahsavari
Working with remote and international clients

Software Engineer · Backend & Full-Stack

I build and ship software end to end — and add AI where it solves a real problem.

Seven years of building and maintaining production software: six years of backend systems and MySQL databases, and since 2025 complete products for paying, international clients. Python/Flask backends, REST APIs, Docker deployments and React front ends, with RAG and LLM agents when they fit.

Based in Kermanshah, Iran / Remote worldwide
Focus: Backend · Full-stack · Applied AI
Currently: AI Engineer at Viraco AI

01 / About

Software that works in production.

I build backend services, databases and full-stack products, and I work independently in remote, distributed settings with international clients, communicating in English.

Since 2025 I also design and deploy retrieval-augmented and agentic AI systems for paying clients. Reliability is the common thread: evaluated retrieval, cited answers, and systems that abstain rather than guess.

My master’s research applied neural networks and image tomography to void-fraction measurement in two-phase flow, with six peer-reviewed papers (2023–2025).

02 / Expertise

01Backend & systems: Python, SQL (MySQL), Flask, REST APIs, Docker, multi-tenant systems.
02Frontend & tools: React, Git/GitHub, Claude Code.
03LLM & AI engineering: RAG, CAG, LangChain, vector databases, chunking strategies, reranking, Model Context Protocol (MCP), prompt engineering, NLP, evaluation of LLM systems.
04Machine learning & vision: TensorFlow, OpenCV, computer vision, feature extraction, neural networks (MLP), signal processing.

03 / Selected Work

Project 01 — Open source

protocol-docs-rag

An evaluated RAG assistant for protocol documentation. Answers cite their sources and the system abstains when it cannot verify an answer. Measured hallucination rate: 0.8% on 200 questions from the Uniswap docs.

Python · RAG · Evaluation
View on GitHub ↗

Project 02 — Open source

dao-treasury-analyst

A verifiable AI analyst for DAO treasuries. Every reported figure traces back to its on-chain transactions, and the system flags figures it cannot verify.

Python · On-chain data · LLMs
View on GitHub ↗

Project 03 — Open source

AI-Native Website Agent Layer

Turns any website into a trustworthy, actionable source for AI agents, with tiered permissions and API keys over REST and MCP, plus content-update detection.

REST · MCP · Agents
View on GitHub ↗

Project 04

cag-drug-explorer

Compares RAG and CAG retrieval architectures on token budget, cache hit rate and latency, using live openFDA drug-label data.

Python · RAG · CAG · openFDA
View on GitHub ↗

Project 05

nobatyar

A clinic management system covering scheduling and patient management.

Flask · React
View on GitHub ↗

Project 06

my-coding-agent

A coding agent for rapidly building MVPs.

LLM agents
View on GitHub ↗
All repositories on GitHub ↗

04 / Experience

AI Engineer · Full-Stack — Viraco AI

Kermanshah, Iran · Aug 2025 – Present

  • Built and deployed full-stack products for paying clients, from backend and database to deployment.
  • Royan Darou: a veterinary/pharmaceutical assistant that answers product, dosage and alternative-medication questions for veterinarians.
  • Kermanshah University of Technology: a QA assistant for educational questions, built to prevent incorrect answers.
  • QR-code product manual assistant: customers scan a product and query its manual in natural language.
  • Worked with multi-tenant architecture, REST and MCP integrations, vector databases, reranking and retrieval evaluation.

Backend Developer — Ferosha

Kermanshah, Iran · Jul 2019 – Aug 2025

  • Built and maintained backend services and MySQL databases over six years.

Education

Kermanshah University of Technology

  • M.Sc. Electrical Engineering (Electronics) · 2020 – 2023 · GPA 17.87 / 20
  • B.Sc. Electrical Engineering (Control) · 2016 – 2020

Publications & courses

  • Six peer-reviewed papers (2023–2025) applying neural networks and image tomography to capacitance-sensor measurement of void fraction in two-phase flow. Google Scholar ↗
  • Kaggle: Pandas, Feature Engineering, Data Cleaning (2024) · Artificial Intelligence and Computer Vision, Robotics and Mechatronics Union of Iran (2019) · Project: flying-object detection in the sky with R-CNN (2024)

05 / Contact

Have an interesting problem to solve?