Data Scientist

📋 Description

• • Invent the future of flight with data. As a Data Scientist at GE Aerospace in Alpharetta, you will join a cross-disciplinary squad that turns terabytes of aviation data into safer, more efficient skies. Your models will predict engine-health trends before they ground a fleet, optimize flight paths to save millions in fuel, and power computer-vision systems that inspect turbine blades in seconds instead of hours.
• • Own the full ML life-cycle. From framing the business question with product managers to deploying containerized models on AWS SageMaker or Azure ML, you will design, train, validate, and monitor algorithms that run at cloud scale and on the edge inside aircraft. You will write production-grade Python, version-control experiments in Git, and automate retraining pipelines using Airflow or GitHub Actions.
• • Champion Retrieval-Augmented Generation. You will build RAG pipelines that let maintenance crews ask natural-language questions—“Why did engine 2-3 surge at FL350?”—and receive evidence-backed answers synthesized from decades of logbooks, bulletins, and IoT telemetry. Expect to fine-tune open-source LLMs, chunk unstructured text, and index vectors in Pinecone or FAISS for sub-second retrieval.
• • Dive into time-series forecasting & anomaly detection. Using millions of flight-hours of sensor data, you will develop Prophet, ARIMA, and deep-learning models (LSTM, Transformer) that forecast part-life remaining and flag anomalous vibrations 50 flight cycles before traditional thresholds. Your work directly reduces unplanned removals and keeps aircraft where they belong—in the air.
• • Drive root-cause analysis at fleet scale. When a new fault code appears across 500 aircraft, you will apply Bayesian networks, SHAP values, and causal graphs to isolate whether the culprit is a software revision, a batch of fuel nozzles, or an environmental factor. Insights are packaged in interactive dashboards (Streamlit, Plotly Dash) for engineers and FAA regulators.
• • Collaborate in Data Science Workouts—our internal hack-sprints. Bring a hypothesis on Monday, demo a working prototype on Friday, and see it funded for fleet-wide rollout. Past workouts cut single-engine taxi fuel burn by 2 % and generated two patent filings.
• • Ensure iron-clad data quality. Partner with data engineers to build Great Expectations suites, dbt transformations, and anomaly detectors that catch sensor drift before it poisons models. You will also create synthetic data to augment rare failure modes and satisfy export-control restrictions.
• • Communicate like a pilot briefs a crew. Translate ROC curves, MAPE scores, and confusion matrices into concise, jargon-free narratives for VPs, mechanics, and airline customers. Publish internal blogs, present at brown-bags, and represent GE Aerospace at Kaggle Days and ODSC.
• • Document for regulators. Generate model cards, bias audits, and FAA-compliant validation reports so every algorithm that influences flight safety can withstand external review. Archive reproducible artifacts in MLflow and secure artifact repositories.
• • Grow with us. Expect 10 % time for self-directed learning, access to 200 TB of aviation data, and mentorship from staff data scientists who have filed 30+ patents. Whether you want to become a deep-learning researcher, an AI product manager, or a technical fellow, we will build the flight plan together.