diff --git a/TODO_PIPELINES/samantha_pipeline.md b/TODO_PIPELINES/samantha_pipeline.md
new file mode 100644
index 0000000..6227fec
--- /dev/null
+++ b/TODO_PIPELINES/samantha_pipeline.md
@@ -0,0 +1,58 @@
+**Sport API**
+
+## API sports
+site: https://api-sports.io/
+- cover multiple sports
+
+## sports data io
+site: https://sportsdata.io/
+- mainly NFL, NBA, MLB, NHL and others
+
+## ThesportsDB
+site: https://www.thesportsdb.com/
+- free JSON sports API for scores, teams, players, and artwork
+
+## RapidAPI
+site: https://rapidapi.com/user/api-sports
+
+
+**Finance APIs**
+
+## Alpha Vantage
+site: https://www.alphavantage.co  
+- for real time and historical stocks and technical indicators 
+
+## Marketstack
+site: https://marketstack.com  
+- real time and historical EOD price data for many global exchanges
+
+
+**Workflow**
+1. data integration
+- pull data from sports API and save the raw data into storage folder
+
+2. clean and organize
+- use pandas to clean and standardize values
+- create table to save those processed data into data warehouse
+
+- Postgres/BigQuery/Snowflake
+
+3. data preparation
+- split data into training, validation, and test sets
+
+4. build and train the model
+- use PyTorch to build a simple transformer model 
+- train model using deep learning framworkm and predict the target output
+
+5. evaluate
+- look at error metrics and some plots to see if predictions look reasonable by using the data outside of training
+- analyze the errors and refine the model
+
+6. prediction
+- package the final model into an inference service
+
+- FastAPI
+
+7. monitoring and maintenance
+- log predictions and track real-world performance over time
+- retraining workflows if drift or degradation is detected.