This case study describes a large-scale AI transformation at an automotive company focused on improving the speed, reliability, and usability of enterprise data within a complex global IT environment.
Through a phased roadmap, we modernized the Snowflake data platform, introduced natural language analytics using Generative AI on AWS Bedrock and LangChain, and laid the foundation for ML-driven anomaly detection to support more proactive customer operations
Meet Our Client
The client is a global brand delivering high-volume digital services for the automotive brands. These services support remote vehicle functions, real-time telemetry from millions of connected vehicles, and mobile applications used daily by drivers.
Case Study Shortcut
Challenge
Data Infrastructure Optimization
Moved heavy processing from Tableau into Snowflake by implementing optimized data models and pre-aggregated views. This reduced dashboard refresh times from hours to seconds and established a consistent data foundation for reporting and AI use cases.
AI-Powered Data Access
Introduced anomaly detection to identify potential vehicle issues earlier and shift operations toward proactive service. Automated model retraining will follow in future phases.
Proactive Operations Enablement
Implemented an anomaly detection service that enables earlier identification of potential vehicle issues, shifting support from reactive to proactive operations. Automated model retraining capabilities will be added in future phases.
Goal
The program aimed to modernize the data ecosystem to support faster decision-making, scalable analytics, and future AI capabilities while operating within strict enterprise constraints.
- Improve executive visibility by removing dashboard performance bottlenecks and optimizing data processing in Snowflake
- Build scalable pipelines to ingest and structure telemetry data from millions of connected vehicles
- Enable self-service analytics for non-technical users through natural language interfaces with strong governance
- Support proactive service through automated detection and alerting
- Strengthen delivery reliability through containerization, SSO integration, and CI/CD practices
- Maintain reporting continuity during the migration from Tableau to Power BI
Outcome
The engagement enabled a shift from slow, reactive reporting to faster, AI-enabled decision-making. It improved data reliability, accelerated insight delivery, and created a foundation for continued platform evolution. Close coordination across globally distributed teams ensured steady progress while maintaining operational stability
Before
- Updated once daily; took ~1 hour to process
- Performed in Tableau UI layer pulling raw datasets
- Tableau extracting millions of rows for client-side aggregation
- Required SQL proficiency and direct database access
- Reactive; manual incident response after customer calls
After
- Sub-second query responses with near real-time data visibility
- Pre-computed in Snowflake materialized views with incremental refresh
- Optimized Snowflake queries returning only aggregated results
- Natural language queries via LLM-powered chatbots with semantic understanding
- Proactive; automated anomaly detection with pre-emptive alerting
Integrate those solutions in your company
Case Study Details
Approach
Scalable Data Architecture
- Dashboard latency was caused by heavy client-side processing. We shifted business logic into Snowflake using optimized SQL and materialized views, leveraging parallel processing and reducing data transfer overhead.
Telemetry Data Pipelines
- Implemented continuous ingestion and transformation pipelines handling millions of telemetry events daily with built-in validation and normalization.
Natural Language Analytics
- Built an LLM-powered interface with a semantic layer that translates user questions into accurate Snowflake queries.
Anomaly Detection and Learning
- Developed automated detection and alerting with feedback-driven retraining to continuously improve accuracy.
Unified Data Foundation
- Both dashboards and AI services now consume the same curated datasets, ensuring consistency and simplifying maintenance.
Platform Migration
- Migrated reporting from Tableau to Power BI while preserving performance and business logic.
Enterprise Deployment Model
- Delivered detailed deployment runbooks and supported manual releases through structured validation and rollback processes.
Technology
AWS
SageMaker
React
SecureAuth
Jenkins
Our team
Jakub Chmielewski
Senior Data Engineer
Michał Pocztowski
Senior Data Scientist
Mikołaj Sitarz
Principal Data Scientist
Maciej Trzaskalski
Project Manager
Piotr Danielczyk
Senior Data Engineer
Our Team Expert Opinion
On this project, about 30% of the work was deep engineering and 70% was navigating enterprise complexity. We ran the engagement as a long-term roadmap, with each initiative building on the last while working within highly manual processes. The technical scope included optimizing Snowflake, building LLM-powered semantic layers, and deploying anomaly detection pipelines. The real win was making advanced engineering succeed inside rigid enterprise delivery models, bridging cutting-edge technology with large-scale collaboration.
Maciej Trzaskalski Project Manager – Addepto
