Experience
Head of Data & Automation — Sotoodeh Automotive Group
- Led a 5-engineer team; owned architecture, delivery, and operational reliability.
- Deployed a 10-node N8N automation cluster (~54K executions/min) and a FastAPI webhook engine (~146K API calls/min).
- Built LLM-powered agentic systems (LangChain) for data cleanup, document validation, and sales assistance.
Senior System Analyst — Snapp Box
- Rebuilt critical analytics layers on ClickHouse for a major logistics feature.
- Designed monitoring and business dashboards (Power BI).
AI & Full Stack Developer — MATA
- Built AI agents for OCR, data extraction, and multi-agent automation.
- Developed FastAPI backends with Pydantic-validated schemas and LLM generators.
Python Developer → Python Team Lead — Airmee
- Owned core allocation microservices; cut latency and improved stability.
- Integrated ML for ETA prediction, load forecasting, and route-feasibility scoring (AWS Lambda).
Core Skills
ML & Statistics
Forecasting, recommenders, A/B testing, survival analysis — scikit-learn, XGBoost, PyMC
Deep Learning & NLP
Transformers, sequence models — Keras, TensorFlow
LLM Agents
Multi-agent orchestration, retrieval — LangChain, PydanticAI, OpenAI/Gemini SDK
RL & Optimization
Bandits, constrained RL, LP/MILP — TF-Agents, Ray RLlib, OR-Tools
MLOps & Data
Docker, FastAPI, MLflow, Airflow, AWS, ClickHouse, PostgreSQL
Education & Awards
- M.Sc., Sharif University of Technology (2017–2019)
- B.Sc., University of Isfahan (2009–2013)
- 2nd Place — Amirkabir AI Competition (encoder–decoder forecasting) · Winner — AIII Movie Recommender Competition (two-tower recommender)
- RL Specialization — Amii · ML/DL Specializations — Coursera
Contact
For job opportunities or project inquiries, send me a message — I typically reply within 48 hours.
LinkedIn: amirhoseinmotaharpour