Amirreza Madani

About Me

I work end-to-end as an ML Engineer / Data Scientist, from research to shipped, containerized services, with recent focus on LLM & RAG systems, multimodal industrial sensor fusion (thermal/visual imaging, vibration analysis), and genomics data analysis.

I started my career as an Android and Flutter developer, then moved into Machine Learning and Data Science — building on a B.Sc. in Electrical Engineering (Control) from Sharif University of Technology and a National Physics Olympiad gold medal.

Skills

Languages: Python, Java, Kotlin, Dart, Scala, Haskell, SQL
ML/DS: PyTorch, TensorFlow, JAX, scikit-learn, Pandas, NumPy, SciPy
LLM & RAG: Embeddings, vector databases (ChromaDB), local LLM serving (Ollama), prompt engineering
Backend: FastAPI, Spring Boot, REST APIs
DevOps: Docker, Kubernetes, Docker Swarm, GitLab CI, GitHub Actions
Mobile: Native Android (Java/Kotlin), Flutter/Dart, TensorFlow Lite

Experience

Machine Learning Engineer — WreckAugury

2025 - 2026

Built a RAG-based AI assistant: multilingual embedding model evaluation, chunking and retrieval pipeline (ChromaDB + Ollama), and a FastAPI service deployed with Docker.
Built ML pipelines for automated thermal/visual industrial defect analysis, including a custom clustering algorithm and image-segmentation methods.

Java & Python Developer — FootFoot

2025

Built a Spring Boot API server and Python bots (Telegram/SPlus) for a Persian-language football data assistant.
Deployed the full stack on a two-node Docker Swarm cluster with Nginx and GitLab CI/CD.

Python Developer — HumanGen

2023 - 2024

Analyzed large-scale DNA data to support medical diagnosis.
Built data-scraping, merging, and graph-analysis pipelines for genetics research.

Flutter Developer — 6thSolution Technologies / REKAB

2021 - 2022

Contributed to a cross-platform mobility app (REKAB → City).
Created the open-source dart_sealed package to modernize the codebase.

Android Developer — Bato Intelligent Personal Assistant

2019 · 10 months

Built core modules — hotword detection, voice recognition, cryptography (AES/RSA/MAC across multiple API levels), and a transparent REST authentication system — for a natural-language personal-assistant Android app.
Shipped to 5k+ users with a 98% crash-free rate; used Java, Kotlin, Dagger, RxJava, Room, and Retrofit.

Projects

shamsi_date Dart and Flutter Package - A Flutter and Dart package for using Jalali (Shamsi) calendar. Enables conversion, formatting, and manipulation of Jalali and Gregorian dates and times. Pub pages: shamsi_date
dart_sealed Dart and Flutter Package - A Dart and Flutter library for generating sealed classes with abstract super types and data subclasses. This project is mostly developed by me. There is a fixed and updated version of this package here, waiting to be merged to upstream repository. Pub pages: sealed_annotations, sealed_generators, sealed_writer,
harmony_log Dart and Flutter Package - Harmony Logging Mechanism. It has various implementations of filters and outputs. Unfortunately the GitHub repository is private, but you can see the source code by downloading it from Pub. I also made a clone of project in GitLab with fixes and upgrades for Dart 3 here, but keep in mind that the code is proprietary. Pub pages: harmony_log
harmony_auth Flutter Package - Harmony Low-Level Authentication and Token Management Mechanism. Unfortunately the GitHub repository is private, but you can see the source code by downloading it from Pub. I also made a clone of project in GitLab with fixes and upgrades for Dart 3 here, but keep in mind that the code is proprietary. Pub pages: harmony_log