ANDREW ELLIOTT
SENIOR AI/ML ENGINEER DATA SCIENTIST
Seattle, WA • 206-***-**** • **.**********@*****.*** PROFESSIONAL SUMMARY
Senior AI/ML Engineer with 12+ years of experience in machine learning, data science, and software engineering across healthcare and large-scale technology platforms. Skilled in Python, PyTorch, TensorFlow, scikit-learn, SQL, Apache Spark, and cloud-based ML infrastructure, with expertise in predictive modeling, recommendation systems, deep learning, and MLOps. Experienced in developing scalable AI solutions, including LLMs, Generative AI, RAG, and production ML pipelines, while collaborating with cross-functional teams to deliver reliable, data-driven applications. TECHNICAL SKILLS
Programming Languages: Python, SQL, Java, Bash, JavaScript Machine Learning & AI: scikit-learn, XGBoost, LightGBM, PyTorch, TensorFlow, Keras, Supervised Learning, Unsupervised Learning, Classification, Regression, Ensemble Methods, Feature Engineering, Hyperparameter Tuning, Predictive Modeling, Anomaly Detection Deep Learning & Recommendation Systems: Neural Networks, Deep Learning, Embedding Models, Transformers, Attention Mechanisms, Recommendation Systems, Learning to Rank, Time-Series Forecasting, Distributed Model Training Generative AI & NLP: Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), Hugging Face Transformers, LangChain, LlamaIndex, Prompt Engineering, Vector Embeddings, Semantic Search, NLP, NLTK, spaCy, LLM Evaluation Data Science & Analytics: Pandas, NumPy, SciPy, Statsmodels, Matplotlib, Seaborn, Exploratory Data Analysis (EDA), Statistical Modeling, Hypothesis Testing, A/B Testing, Experiment Design, Model Evaluation, Causal Analysis Big Data & Data Engineering: Apache Spark, PySpark, Apache Kafka, Apache Hive, Apache Airflow, ETL/ELT Pipelines, Batch Processing, Stream Processing, Data Validation, Data Transformation, Data Quality Management MLOps & ML Infrastructure: Uber Michelangelo, MLflow, Docker, Kubernetes, Model Serving, Model Monitoring, Model Versioning, CI/CD, Automated Training Pipelines, Data Drift Detection, Experiment Tracking, Model Retraining Cloud & Backend Technologies: AWS, Google Cloud Platform (GCP), REST APIs, FastAPI, Flask, Microservices, Distributed Systems, Git, Linux, Automated Testing
Databases & Data Storage: PostgreSQL, MySQL, Redis, Apache Hive, Relational Databases, Data Warehousing, Vector Databases Healthcare & Domain Expertise: Healthcare Data Analytics, Electronic Health Records (EHR), Clinical Data Integration, Patient Risk Analysis, HIPAA-Aware Data Processing, Telemedicine Analytics, Marketplace Optimization, Delivery Time Prediction PROFESSIONAL EXPERIENCE
UBER Mar 2020 – Present
AI/ML Engineer & Data Science
PROJECTS: Uber Eats Personalization & Recommendation Systems • Delivery Time Prediction • Michelangelo ML Platform • Internal Generative AI Applications
• Contributed to the design, development, and production deployment of AI/ML solutions supporting Uber Eats personalization, restaurant recommendation systems, delivery time prediction, and internal AI-powered productivity applications. Developed scalable machine learning pipelines and predictive services that supported personalized customer experiences, operational forecasting, and data-driven decision-making across high-volume marketplace workflows.
• Designed and developed end-to-end machine learning pipelines using Python, SQL, PySpark, Apache Spark, and Hive to prepare order, delivery, and customer interaction data for training. Optimized partitioning and transformations, cutting a recurring batch pipeline's median runtime by 22% based on Spark job-history comparisons.
• Built and optimized predictive machine learning models using Python, scikit-learn, XGBoost, and gradient boosting for delivery time estimation and demand forecasting. Engineered temporal and geographic features and evaluated held-out predictions, reducing ETA mean absolute error (MAE) by 8% relative to the previous model baseline.
• Developed and enhanced personalization and recommendation models using PyTorch, TensorFlow, embeddings, and ranking features for restaurant discovery. Tested candidate-ranking changes on offline datasets, improving NDCG@10 by 5% relative to the prior ranking model.
• Engineered scalable data processing and feature generation workflows using Apache Spark, PySpark, Kafka, SQL, and Michelangelo feature infrastructure for batch and near-real-time ML. Improved feature validation and transformation efficiency, reducing median feature-availability lag by 28% in pipeline monitoring.
• Performed advanced data analysis, statistical experimentation, and model evaluation using Python, Pandas, NumPy, SciPy, and SQL. Designed offline evaluations and A/B testing analyses, applied hypothesis testing and confidence intervals, and assessed model behavior using metrics including ROC-AUC, precision, recall, ranking quality, and forecasting error to support model selection and product decisions.
• Integrated and deployed production machine learning models through Uber's Michelangelo platform using Python, REST APIs, Docker, and Kubernetes. Optimized model-serving paths and request payloads, reducing p95 inference latency from 135 ms to 112 ms (17%) in service telemetry while maintaining validation and rollback controls.
• Optimized deep learning training and inference workflows using PyTorch, TensorFlow, distributed data processing, and GPU-enabled computing, working on model architecture evaluation, data batching, computational efficiency, and inference performance. Conducted systematic experiments to balance prediction quality, model complexity, training cost, and production latency requirements.
• Developed Generative AI and LLM-powered applications for internal engineering knowledge retrieval beginning in 2023, using Python, Hugging Face Transformers, embeddings, semantic search, and RAG. Tuned document chunking and retrieval evaluation, improving top-5 evidence retrieval recall from 0.73 to 0.81 on a labeled evaluation set.
• Built backend services for machine learning and LLM-based applications using Python, FastAPI, REST APIs, and enterprise model endpoints. Added asynchronous handling, caching, structured logs, retrieval evaluation, and sensitive-data safeguards, lowering p95 application response time from 2.1 s to 1.8 s (14%) in service traces.
• Supported MLOps, production monitoring, and model lifecycle management, implementing automated validation, deployment checks, performance tracking, data quality controls, and model health monitoring. Investigated prediction degradation, data distribution changes, and production inference issues while collaborating with infrastructure teams on CI/ CD workflows, containerized services, and model retraining processes.
• Collaborated closely with data scientists, machine learning engineers, backend developers, product managers, and infrastructure teams to define technical requirements, evaluate modeling approaches, and deliver reliable AI capabilities. Mentored junior engineers, conducted code reviews, contributed to architecture discussions, documented technical decisions, communicated model limitations, and promoted engineering best practices across cross-functional initiatives. THIRD EYE HEALTH Jun 2014 – Feb 2020
Software Engineer Healthcare Data Science & Machine Learning PROJECTS: Telemedicine Platform • Clinical Data Analytics • Electronic Health Record Integration • Patient Risk Analysis • Healthcare Reporting
• Contributed to the development of healthcare data engineering, clinical analytics, and predictive modeling capabilities supporting Third Eye Health's telemedicine platform, physician consultation workflows, electronic health record integrations, and clinical reporting systems. Helped build data-driven solutions for analyzing patient care information, monitoring consultation activities, improving reporting consistency, and supporting operational decisions across post-acute care environments.
• Developed healthcare data processing and analytical applications using Python, SQL, Pandas, and NumPy to normalize patient encounters, consultations, and facility data. Added reusable validation and duplicate-detection checks, lowering downstream duplicate-record exceptions by 31% in scheduled data-quality reports.
• Performed clinical data analysis and statistical modeling using Python, Pandas, SciPy, and scikit-learn, examining relationships among patient characteristics, consultation patterns, historical healthcare utilization, and hospital transfer outcomes. Applied descriptive statistics, regression analysis, and exploratory modeling to identify relevant trends and evaluate factors associated with care utilization.
• Designed and maintained ETL pipelines using Python, SQL, PostgreSQL, and scheduled jobs to consolidate telemedicine and EHR-related data. Improved incremental loading, indexing, and query plans, reducing median nightly ETL runtime by 23% according to scheduler logs.
• Developed exploratory data analysis and healthcare reporting workflows with Pandas, NumPy, SciPy, SQL, and visualization libraries to analyze consultations, response patterns, and utilization. Automated recurring extracts and reporting, cutting weekly report preparation from 4 hours to 2.5 hours (38%) based on recurring analyst workflow records.
• Built and maintained Python-based backend services and REST API integrations to exchange patient demographic and consultation information with EHR systems. Implemented mapping validation, retries, and structured error logging, reducing failed field mappings by 18% across integration test runs.
• Engineered data preparation and feature extraction workflows using Python, Pandas, SQL, and scikit-learn to build structured variables from historical encounters and consultations. Standardized missing-value handling and completeness checks, raising completeness of selected reporting fields from 92% to 97% based on SQL profiling queries.
• Applied traditional machine learning and statistical techniques, including logistic regression, decision trees, and random forests, to exploratory healthcare-utilization and patient-risk analysis on historical data. Used cross-validation and feature selection to improve offline ROC-AUC from 0.70 to 0.75 versus a research baseline; no automated clinical decisions were made.
• Conducted analytical model validation using Python, scikit-learn, statistical testing, cross-validation, and classification metrics, reviewing model performance, data limitations, and potential sources of error. Evaluated precision, recall, F1-score, and ROC-AUC where appropriate, documenting findings and assessing whether analytical approaches were suitable for further investigation.
• Developed and maintained SQL reporting queries, analytical datasets, and automated reporting workflows using Python and relational databases to monitor telemedicine consultation activity, physician response times, facility performance indicators, and operational utilization trends. Reduced repetitive manual data preparation through scheduled transformations, reusable queries, and standardized reporting logic.
• Supported the development and maintenance of Python application services using Flask, REST APIs, SQL, Git, and automated testing practices, improving data-processing reliability, query performance, logging, and maintainability. Participated in debugging, integration testing, deployment preparation, and technical documentation while following healthcare data privacy and access-control requirements.
• Collaborated with software engineers, physicians, clinical operations specialists, healthcare administrators, and product stakeholders to understand healthcare workflows and translate operational requirements into reliable technical solutions. Participated in code reviews, mentored teammates, documented application and data-processing workflows, communicated technical findings to nontechnical stakeholders, and contributed to continuous improvements in engineering quality and team collaboration.
EDUCATION
MICHIGAN STATE UNIVERSITY EAST LANSING, MI Aug 2021 – May 2023 Master of Science in Computer Science
INDIANA UNIVERSITY BLOOMINGTON, IN Aug 2011 – May 2015 Bachelor of Science in Computer Science