A naukri.com initiative
Swiggy
1w
168
Image Credit: Swiggy
Building Rock-Solid ML Systems
- Machine learning plays a crucial role in Swiggy's food delivery operations, from predicting order ETAs to recommending groceries on Instamart.
- Key areas for ML operational excellence at Swiggy include Exploratory Data Analysis (EDA), Sensitivity Analysis, Explainable AI, and Coding Standards.
- EDA involves understanding data distributions, detecting anomalies, and ensuring stability across training and testing environments.
- Sensitivity Analysis helps identify feature influence on model behavior and establish effective operating ranges for features.
- Explainable AI addresses the 'black box' problem by making models understandable through tools like SHAP.
- Coding Standards at Swiggy focus on clean code, smart implementation, thorough review, rigorous unit testing, and clear documentation.
- Through Sensitivity Analysis, feature operating ranges are identified, thresholds established, and caps implemented to manage outliers in production.
- Explainable AI techniques like SHAP help attribute predictions to specific inputs, making complex models understandable.
- Unit Testing in TensorFlow ensures individual components work correctly within data pipelines, enhancing model reliability.
- Adhering to coding standards and documentation ensures the delivery of quality, maintainable ML solutions at Swiggy.
Read Full Article
6 Likes
For uninterrupted reading, download the app