Data science and Internet of Things on the edge
summary
A tiny leak sensor is built and deployed: two temperature sensors on a pipe infer flow from a drop, run fully offline on 32 kilobytes RAM and two AAA batteries, wake only to upload a few times a day. Data collection uses a pipe chamber lab plus long home trials, manual labeling and an internal time‑series viewer that becomes widely used. Models start in Python, get compressed (float64 → small ints) and converted to C with unit tests until a model beats heuristics and survives firmware cycles.
AI-written summary · about
chapters (8)
- 0:00IoT and edge data science intro
- 2:57Edge computing for IoT
- 5:08Edge data science workflow
- 10:05Real-world leak detection data
- 14:39Hardware limits and model simplification
- 19:21Deploying models to devices
- 23:34Edge IoT development lessons
- 25:48Q&A: Elderly monitoring
related talks
39mFrom a Weekend Hack, to a Future-Proof IoT Device
37mFrom Sensors to Insights at the Edge: Applying AI to IoT Data
39mFrom a Weekend Hack, to a Future-Proof IoT Device
52mApplying AI to IoT Data
16mEdge Computing and next-gen of IoT Sensors
32mData Science in the Internet of Things with Python and Spark
01h 01mML and the IoT: Living on the Edge
54mSmarter Everything with ML & the IoT