Julia M.

I build data systems that clean themselves up, report on themselves, and predict what comes next. Over five years in Python and SQL across AWS and Azure, I design the pipelines and warehouses that get data in shape, the Power BI dashboards leaders act on, and the models that forecast where a number is heading and how likely it is to hit its target. My work has improved operational efficiency by up to 30%, cut manual work hours by 80%, and replaced daily analyst work with automation that runs on schedule and only wakes a human when something looks wrong. At Fannie Mae I automated regulated-data ETL across AWS. On Syndication prediction engine project, I built an 8-stage pipeline that reads messy radio play logs from 168 stations every day, separates real signal from noise, and hands a clean result to the reporting team. On MAPLE/PTT, a patent-pending forecasting product, I built the prediction engine plus the layer that makes its numbers defensible: every forecast comes with a real probability and a visible trail back to where that probability came from, which is what survives a governance review. That project earned a 5.0 client rating. - Data Engineering & Warehousing: Snowflake, Databricks, Spark, AWS Glue, Athena, Aurora, PostgreSQL, Azure SQL, Microsoft Fabric, Parquet, schema design, incremental upserts, dedup - ETL & Orchestration: Python ETL pipelines, Step Functions, EventBridge, S3 event triggers, CloudFormation, CloudWatch, SNS alerting - Business Intelligence: Power BI (DAX, Data Modeling, Power Query, KPI Dashboards), automated reporting - Prediction & Forecasting: regression modeling, time-series forecasting, XGBoost, LSTM, likelihood scoring, model calibration, ensemble methods - Pattern & Anomaly Detection: clustering, sequence matching, duplicate detection, confidence scoring, outlier flagging - AI & ML Tools: PyTorch, TensorFlow, Keras, XGBoost, Redis, Llama 3.1, Hugging Face - Languages & Frameworks: Python, FastAPI, Flask, Django, SQL, JavaScript - Data Quality & Testing: format and time zone validation, schema checks, mocked test coverage - Languages & Frameworks: Python, SQL, FastAPI, Flask, Django, JavaScript Most clients come to me with data that technically exists but nobody trusts. I make it trustworthy first, then make it useful, then make it tell you something you did not already know. If you need clean pipelines and reporting you can rely on, or predictions on top of them, I would be glad to talk.