Sampling and Modelling the World in Real-Time for Mobility
summary
A real-time platform ingests telematics at 10–100 Hz telematics and GPS at 1 Hz, fuses weather, road features and police accident records, and produces trip-level dynamic pricing as a transactionized premium per trip. Pipeline runs on AWS with Kafka, Flink, Kubernetes, S3, Elasticsearch and GraphQL; Flink enables watermarks and out‑of‑order handling that Spark couldn’t. Models detect events (harsh braking, near crashes from accelerometer), infer blackspots in unseen cities comparable to police data, run fatigue detection from circadian patterns, and will extend to autonomous vehicle and cyber risk under an Innovate UK grant.
AI-written summary · about
chapters (8)
- 0:00Intro
- 1:04Underwriting and rate setting is broken.
- 4:16Real-time dynamic risk calculation for Insurance
- 5:26High Level Stack
- 10:34High Level Platform Architecture
- 17:55Road and Weather Risk
- 22:45Near Crash Detection
- 25:25Challenges
related talks
40mBuilding Real-Time Applications: Cyclist Crash Detection
40mSmart cars of tomorrow: real-time driving patterns
22mBuilding a movement data and analytics platform to de-stress customers & save lives
23mMaking Sense of Streaming Sensor Data: How Uber Detects On-trip Car Crashes
30mReal-Time ML services: From ideation to production
51mReal-time Fraud Detection: Challenges and Solutions
25mReal-Time Fraud Detection at Scale
26mOptimal Approaches for Real-Time Machine Learning with Apache Spark on Kubernetes.