Machine Learning
- PyTorch
- TensorFlow
- Scikit-Learn
- XGBoost
- LightGBM
- CatBoost
- Statistical Learning
- Causal Inference
- Optuna
Naples, Italy · Open to ML / AI Engineer roles
Shayan Ekramnia
Machine Learning Engineer / AI Engineer
I build AI systems that make it all the way to production — and keep working once they're there. End-to-end across data engineering, ML, and MLOps, with deep focus on time-series forecasting at scale.
SE
0Years building ML
0Retail branches served
0Daily forecasts shipped
0SKU-level time series
01 — About
Turning business problems into deployed, monitored ML
I'm a Machine Learning Engineer with 5+ years designing, training, and deploying production-grade ML and AI systems end to end — from data engineering and ETL through model development, MLOps, serving, and monitoring.
Right now I'm the end-to-end architect and developer of a complete production AI system for a national retail network — built from scratch and rolling out across 110+ branches, generating 20,000+ probabilistic forecasts daily over a 50M+-row pipeline. My depth spans time-series & probabilistic forecasting (including fine-tuned foundation models), deep learning, multi-agent reinforcement learning, computer vision, and NLP/speech — grounded in advanced statistical and causal analysis.
I hold an M.Sc. in Data Science (110/110 cum laude) from the University of Naples Federico II, earned alongside full-time industry work.
02 — Skills
A full-stack ML toolbox
Time-Series Forecasting
- ARIMA/SARIMA
- ETS
- Prophet
- TFT
- N-BEATS
- DeepAR
- Chronos (foundation models)
- Hierarchical Forecasting
- Probabilistic & Quantile
- Conformal Prediction
- Backtesting
LLMs & Generative AI
- Hugging Face Transformers
- LLM Fine-Tuning (LoRA/PEFT)
- RAG
- Vector Databases
- Embeddings
- Semantic Search
RL & Computer Vision
- PPO
- DQN
- A3C
- Multi-Agent RL
- Gymnasium
- OpenCV
- YOLO
- DeepFace
- CoreML
MLOps & Deployment
- Docker
- Apache Airflow
- MLflow
- Model Registry
- Drift Monitoring
- CI/CD (GitHub Actions)
- FastAPI
- GPU Training
Data & Cloud
- Python
- SQL
- Apache Spark
- Apache Kafka
- ETL/ELT
- PostgreSQL
- Neo4j
- Redis
- AWS
- Linux
03 — Experience
Where I've shipped
Machine Learning Engineer / AI Engineer
Nov 2024 — PresentSIACLOUD (TEKNOTEAM S.R.L.) · Naples, Italy
- Single-handedly designed, built, and deployed a complete production AI system — data engineering, ML engineering, and MLOps — for a retail network of 110+ branches, replacing fully manual, spreadsheet-driven planning.
- Ship 20,000+ probabilistic (quantile) forecasts daily across 7,000+ SKU-level time series, owning ingestion and ETL of a 50M+-row dataset through training, serving, and dashboards.
- Fine-tune time-series foundation models (Chronos-Bolt) alongside classical models, with continuous automated evaluation on rolling held-out windows.
- Perform advanced exploratory and causal analyses of company-wide retail data to surface demand drivers and inform model design and commercial decisions.
- Automated the ML lifecycle with Airflow, MLflow, and Docker: reproducible retraining, versioned registry, one-click rollback, and GPU-accelerated containerized serving.
Machine Learning Engineer & Software Developer
Jun 2018 — Oct 2022Vira Vision Co. · Esfahan, Iran
Selected through a university elite-talent program for placement at a high-tech park company; worked alongside B.Eng. studies until 2020, then full-time.
- Engineered an AI-based COVID-19 risk-prediction system deployed across 18 companies and 28,000 employees during the pandemic.
- Built and deployed ML models for customer-behaviour prediction across the company's SaaS products, informing retention campaigns and product analytics.
- Developed Python SaaS modules (CRM, ERP, supply-chain) with embedded analytics and ML-driven recommendations for enterprise clients.
- Built automated data pipelines, feature-engineering workflows, and BI dashboards (Tableau, Power BI) with post-deployment monitoring and A/B testing.
04 — Projects
Selected work
Multi-Agent RL for Warehouse LGV Optimization
A complete MARL system optimizing routing and task sequencing for automated guided vehicles, with a custom Gymnasium environment (kinematic constraints, collision avoidance, dynamic pallets).
PPO ranked #1 overall — efficiency 78.4/100 vs. 65.3 for A*, 87.5% completion, validated with Welch's t-tests (p<0.001).
- PyTorch
- PPO/DQN/A3C
- Gymnasium
- Streamlit
Real-Time Facial Emotion Detection
A real-time emotion-recognition pipeline (DeepFace + OpenCV) supporting 7-class Ekman and 3-class valence outputs, with an internal HTTP API exposing rolling aggregates.
Benchmarked five detector backends (RetinaFace, MTCNN, MediaPipe, SSD, OpenCV) for accuracy–latency trade-offs.
- DeepFace
- OpenCV
- FastAPI
Voice-Enabled Quote Retrieval Chatbot
An end-to-end voice NLP system over a 1.3M-quote corpus in Neo4j with full-text indexing — integrating Whisper ASR, SpeechBrain speaker ID (192-dim), and TTS for a personalized multi-user experience.
Built fuzzy author resolution and top-K ranked retrieval with intent routing.
- Whisper
- SpeechBrain
- Neo4j
On-Device AI · iOS
Banalize & EyeLight — iOS ML Apps
Banalize: CoreML app assessing fruit ripeness and classifying bread types with privacy-preserving on-device inference. EyeLight: assistive-vision app using CoreML + Vision for real-time object detection, distance estimation, and obstacle alerts for visually impaired users.
- CoreML
- Vision
- Swift
05 — Education & Recognition
Credentials
M.Sc. Data Science
2023 — Oct 2025University of Naples Federico II
Apple Developer Academy Diploma
Sep 2024 — Jun 2025University of Naples Federico II
B.Eng. Mining Engineering
Sep 2015 — Mar 2020Isfahan University of Technology
Awards & Certifications
- ★National Elite — INEFIran's National Elites Foundation, 2020 — national selection of top graduates and young talents.
- ▴Professional Project ManagementQAL UK Academy · Dec 2022
- ▴Business AdministrationUniversity of Tehran · Sep 2023
Languages
- EnglishFluent
- PersianNative
- ItalianIntermediate
06 — Contact
Let's build something that ships.
Open to Machine Learning / AI Engineer roles and interesting applied-AI collaborations. The fastest way to reach me is email.