Junior ML/AI Engineer · Data Analyst · MSc Data Science @ UniTo
I turn raw, messy sensor data into decisions that engineers and managers actually use.

• 🔬 Currently working as Data Analyst & ML Engineer Intern at Eurix — building LSTM Autoencoder pipelines for unsupervised anomaly detection in real IoT/HVAC sensor data
• 🎓 MSc in Stochastics & Data Science, University of Turin — graduating July 2026
• 📊 Thesis: Advanced Data Analysis for HVAC System Characterization and Evaluation of Environmental Comfort and Energy Performance
• 🏆 1st place — Michelin Students Green Challenge | Finalist — UniTo–SKF Indigo Challenge
• 🎓 Former Teaching Assistant in MATLAB @ UniTo (15–20 students/semester)
• 🌍 Based in Turin, Italy | Open to Turin, Milan, remote

Languages & Libraries

Tools & Workflow

ML Focus Areas

Anomaly Detection · LSTM Autoencoders · Time-Series Analysis · Feature Engineering · Optuna Hyperparameter Tuning · IoT/Sensor Data · EDA · Classification

Industry collaboration (Eurix) + MSc Thesis

Built a full production-ready pipeline for unsupervised anomaly detection in real Building Management System (BMS) sensor data from the Luigi Einaudi Campus, Turin.

• Model: LSTM Autoencoder (TensorFlow/Keras) with Optuna hyperparameter tuning
• Pipeline: 13-module config-driven Python architecture — reproducible, modular, scalable
• Data: Multi-sensor Air Handling Unit data (temperature, humidity, airflow) — real-world, messy
• Techniques: EWMA smoothing, stratified time-series splitting, seasonal regime variants (summer/winter), percentile-based thresholding
• Impact: Findings presented to engineers and facility managers; analysis directly informed maintenance decisions

Python TensorFlow/Keras Optuna Pandas NumPy Matplotlib IoT HVAC

End-to-end classification pipeline on customer behavioral data.

• Logistic Regression + Decision Trees · ~78% accuracy
• Full workflow: data validation → EDA → feature engineering → model evaluation (ROC-AUC, confusion matrix)

Python scikit-learn Pandas Matplotlib

Graph analysis on a Facebook-like network dataset.

• Centrality measures, community detection, and network visualization
• Identified influential nodes and cluster structures

Python NetworkX Matplotlib

• Power BI (dashboards and data storytelling)
• MLOps fundamentals (experiment tracking, pipeline deployment)
• Advanced time-series forecasting methods

🇦🇲 Armenian (Native) · 🇬🇧 English (C1) · 🇮🇹 Italian (B2) · 🇮🇷 Persian (Professional)

I'm actively looking for a paid internship in ML Engineering, AI Engineering, or Data Science — starting from June 2026.

📩 varaga.haghoubians@gmail.com 💼 linkedin.com/in/varagahaghoubians