Template: Data Science Projects

Starter memory edits for data science and ML projects.

---

Core Template (5 edits)

Copy and customize these:

1. "Primary language: [Python 3.11/R 4.3/Julia]"
2. "ML Framework: [TensorFlow/PyTorch/Scikit-learn/XGBoost]"
3. "Data storage: [PostgreSQL/Snowflake/S3/BigQuery]"
4. "Environment: [Jupyter/VS Code/RStudio/Databricks]"
5. "Deployment: [AWS SageMaker/Azure ML/Google Vertex/Docker]"

---

Filled Example

1. "Primary language: Python 3.11, no R" (37 chars)
2. "ML Framework: PyTorch 2.0, TensorFlow only for legacy" (55 chars)
3. "Data storage: Snowflake warehouse + S3 for raw files" (53 chars)
4. "Environment: VS Code with Jupyter extension, not Databricks" (60 chars)
5. "Deployment: Docker containers on AWS ECS, not SageMaker" (56 chars)

---

Extended Template (10 edits)

For more complex projects:

6. "Feature store: [Feast/Tecton/custom]"
7. "Experiment tracking: [MLflow/Weights&Biases/Neptune]"
8. "Data versioning: [DVC/Delta Lake]"
9. "Pipeline orchestration: [Airflow/Prefect/Dagster]"
10. "Model format: [ONNX/TorchScript/SavedModel]"

---

Domain-Specific Additions

NLP Projects

"NLP: Hugging Face Transformers, not spaCy for main models" (59 chars)
"Embeddings: OpenAI ada-002 for production, local for dev" (57 chars)

Computer Vision

"CV Framework: PyTorch + torchvision, not TensorFlow" (52 chars)
"Image storage: S3 with CloudFront CDN" (37 chars)

Time Series

"Time series: Prophet for baseline, custom LSTM for production" (63 chars)
"Frequency: Daily data, no sub-hourly granularity" (49 chars)

---

How to Use

1. Copy the core template
2. Fill in the brackets with your specifics
3. Add using: memory_user_edits add control="..."
4. Test with relevant questions
5. Refine based on results

---

Template from Claude Memory User Edits Guide — CC BY 4.0