Data Scientist

Uniti

About The Position

The Kinetic Data Science team sits within the Customer Success division of Business Operations and builds predictive models that support strategic decision-making across Uniti Solutions’ consumer and business lines. We work with large-scale telecom datasets spanning billing, call center, network, and CRM systems, turning complex enterprise data into actionable predictions, customer segmentation, and model-driven insights. The team is small, collaborative, and moving fast. You will contribute directly to models that influence business decisions. Statistical rigor and clear communication are what we value most, and we welcome strong analytical thinkers from any academic background and what matters to us is the depth of your reasoning, not the discipline of your degree. We are looking for a machine learning engineer with deep statistical foundations, hands-on modeling experience, and an investigative mindset to join the team. You will assist in building, validating, maintaining, and improving predictive models across a range of business domains — customer retention, network performance, marketing, sales, and others as needs evolve. You will develop features from complex multi-source data and help maintain inherited models built by external partners. You will contribute to all phases of the modeling lifecycle, from data exploration through model delivery, under the direction of the team manager. Statistical reasoning and clear communication are the heart of this role. You will spend significant time choosing the right test, validating model assumptions, and quantifying uncertainty. Equally significant time explaining those choices in writing and in person to technical peers and non-technical stakeholders alike. You will report directly to the team manager and work alongside data engineers and solutions architects.

Requirements

2–3 years of experience in a data science or applied statistics role (less experience considered for strong candidates)
Strong foundation in statistical modeling — linear and logistic regression, classification methods, probability theory, and bias-variance tradeoffs, demonstrated through formal coursework, certifications, applied work, or rigorous self-directed study
Working knowledge of applied inferential statistics — parametric and non-parametric hypothesis testing, experimental design (A/B testing, sample sizing, power analysis)
Proficiency in Python for data science (pandas, scikit-learn, numpy, matplotlib/seaborn)
Strong SQL skills, particularly with Snowflake or similar cloud data warehouses
Experience with feature engineering from real-world, imperfect enterprise data — not just clean Kaggle datasets
Ability to work independently and manage your own priorities with minimal oversight
An investigative mindset — you ask why before how, push on assumptions, and follow data anomalies to root causes rather than papering over them
Clear written and verbal communication — you can explain a modeling decision to a non-technical stakeholder and document your work so others can follow it
Bachelor’s degree in any field, or equivalent experience — a STEM degree is not required, and candidates with strong statistical preparation from any academic background are encouraged to apply

Nice To Haves

Experience with Snowpark (Python or SQL)
Exposure to Azure ML or similar cloud ML platforms
Familiarity with MLOps concepts (model versioning, pipeline automation, drift monitoring)
Telecom or subscription-based industry experience
Experience inheriting and maintaining models built by others
Familiarity with Git-based workflows and version control for data science artifacts

Responsibilities

Build, validate, and evaluate predictive models (logistic regression, XGBoost, ensemble methods) across customer retention, network, marketing, sales, and other business domains
Apply statistical reasoning to validate model assumptions, test for confounders, and quantify uncertainty in model outputs
Engineer features from complex, multi-source enterprise data (billing systems, call center logs, CRM, network data) in Snowflake and Oracle
Profile and investigate data quality issues — identify leakage, missingness patterns, join inconsistencies, and source-of-truth conflicts
Maintain and improve inherited production models, including models built in Snowpark by external partners
Perform SHAP-based model interpretability analysis and translate results into business-actionable insights
Design and execute customer segmentation using clustering techniques on model outputs
Write clear, thorough documentation of model logic, feature rationale, data assumptions, and known limitations
Collaborate with the team to define target variables, population filters, and prediction windows grounded in statistical reasoning