Albert Levin

Leading the design and implementation of scalable data and MLOps architectures for enterprise clients.

• Optimized ML pre-processing workloads with an event-driven architecture on Azure with Databricks, PySpark, and EventHub, reducing pipeline runtimes from 2.5 hours to less than 1 hour.
• Built and hosted ML & GenAI apps with LangChain, LangGraph, and React/Node.js.
• Implemented robust CI/CD pipelines with automated testing in Azure DevOps for Python packages, enforcing software engineering best practices and improving team velocity.
• Developed a custom VSCode plugin for local Python/PySpark job debugging, reducing development cycles for the whole team.
• Built a custom monitoring & alerting solution with Python & Grafana on Azure Cloud.

Developed and productionized ML models for forecasting, pricing, and personalization.

• Training & fine-tuning ML models with xgboost, scikit-learn, Tensorflow/Keras.
• Productionizing and serving ML models (Forecasting, Fraud Detection, Recommendation, Search) with FastAPI, Docker, and Helm on Kubernetes in Google Cloud.
• Setting up PySpark pipelines for ML training workloads and Feature Stores in Databricks and MLFlow.

Developed data integration workflows and reporting pipelines for European operations.

• Developed of ETL/ELT pipelines with Python, Airflow, and BigQuery.
• Wrote of web apps with Django and Flask for internal use.
• Set up data modelling in the DWH (Data Vault 2.0).
• Created CI/CD pipelines (GitLab functions) and managed the Airflow Kubernetes cluster.
• Managed security and networks on Google Cloud (firewall, virtual private cloud, load balancer).