Granit Grant Zymberi

About Granit Grant Zymberi

Data Scientist – AI/ML Engineer with a proven track record of designing, deploying, and scaling data-driven and machine learning solutions that bridge analytics, engineering, and strategy. I specialize in building production-ready NLP pipelines, implementing MLOps best practices, and developing end-to-end ETL and predictive workflows that power real-time insights and decision-making.My expertise spans Python, Scala, Spark, Snowflake, BigQuery, Airflow, Terraform, Docker, Kubernetes, and MLflow—enabling me to deliver high-performance systems that integrate seamlessly within enterprise environments. From zero-shot text classification with Hugging Face to optimizing distributed data processing in Spark, I focus on creating architectures that are reliable, interpretable, and scalable.With experience across AI engineering, data engineering, and applied analytics, I bring a full-stack perspective to the ML lifecycle—ensuring models are not only technically sound but also analytically meaningful and business-aligned.Core Competencies: AI/ML Engineering · Predictive Analytics · Natural Language Processing · MLOps · Data Engineering · Cloud Data Platforms · ETL Development · Distributed Computing · CI/CD · Microservices Architecture · REST APIs

Experience

1. Ai Ml Engineer

   SDG Group USA

   Apr 2025 — Present · US

   Lead the design and deployment of machine learning systems that transform raw operational and textual data into predictive intelligence for sales, marketing, and strategic decision-making-Develop clustering and predictive frameworks to forecast client activity and uncover revenue opportunities — leveraging KMeans and time-based models to reveal behavioral patterns and lifecycle transitions-Engineer feature pipelines in Python (pandas, NumPy, scikit-learn) to convert large-scale data into model-ready inputs, ensuring reliability, explainability, and reproducibility in production environments-Design modular NLP pipelines to classify and extract intelligence from unlabeled news data — applying unsupervised topic modeling (LDA, BERT embeddings, K-Means) and Named Entity Recognition (spaCy, Hugging Face) to identify latent themes and contextual entities-Leverage Airflow to orchestrate and monitor ML workflows, enabling traceable, modular automation for daily scoring, retraining, and data backfills-Employ MLflow for experiment tracking, model governance, and metric observability — ensuring a transparent, reproducible lifecycle for training and inference-Containerize ML components with Docker to maintain consistent, portable environments and streamline CI/CD deployment across development and production-Utilize Snowflake and PostgreSQL for scalable model output storage and seamless integration with BI dashboards (e.g, Duomo), empowering business teams to act on real-time predictive KPIs-Create visualization layers (topic maps, clustering insights, performance dashboards) to interpret model results and communicate findings effectively-Collaborate with BI and strategy teams to translate analytical logic into production-grade ML systems that deliver measurable business impact.

Education

• Flatiron School

  Computer Software Engineering

• University of Prishtina

  Bachelor of Science - BS, Computer Science

• UNI- Universum International College

  Bachelor of Science - BS, Business Management