Client Challenges
- Lack of Data Science and ML Engineering expertise with the core team of the Client
- Lack of experience in deployment of ML solutions as integrated parts of the complex Web-based platforms to work under high real-time load
- The market operations teams of the Client in several country markets in LATAM needed a better advisory on the adequate car pricing to support (1) more attractive offer pricing to private car owners, and (2) better margin-driven pricing with their customers to sell the used car in B2B model
Services
- Implemented the used car price forecasting engine for several markets in Latin America where the Client operates
- Provided advanced data analytics on the key price drivers in the customer operational markets
- Implemented data extraction from the corporate BI DWH
- Implemented the data preprocessing and feature engineering pipelines
- Implemented ML models GBDT: (xgboost, lightgbm, catboost; meta-model ensembling)
- Implemented Web API Wrappers around the ML models to make it ready for the customer product integration
Client: E-commerce platform operator (used car sales, operated in 10 countries)
Segment: ML, AI, Data Science
Users base: Hundreds of call center agents in the client facilities in MX
Technology stack: GCP, Python, SQL, scikit-learn, GBDT algos
Team structure: 1 Sr ML Engineer, 1 Middle Data Scientist, 1 Jr Data Scientist