A generic application platform for cloud computing, micro-services, Bigdata and AI.
Million-level event-driven streaming processing for realtime stream processing, joining, analysis, comparison in a unified framework.
Flexible configuration in building Bigdata infrastructure, and quickly develop and deploy application solutions according to customer requirements.
Helium+ IoT Solution
The new IoT solution adopts cloud computing and big data technology architecture, supports the connection of hundreds of millions of IoT devices and the high concurrent processing of trillions of messages, and collects data generated by IoT gateways. The engine performs real-time processing and AI intelligence on the collected IoT data and the results can be integrated with statistics, reporting, searching, monitoring, and alertion in a streaming manner.
The solution is generic enough and has a wide range of usage scenarios, including manufacturing, energy, logistics, wearables, retail, smart buildings, car networking, 3D printing, smart environmental protection and other fields. Most of our use cases are designed and implemented based on this framework, which verify that it can effectively support the general Bigdata and AI requirements.
Helium+ Log Analysis and Monitoring System
The system accepts massive status, logs and other indicators related to system robustness / security and others, The new indicator analysis system returns the massive status, indicators, logs and other series of system robustness and security related indicators returned by other systems. Through the use of the big data real-time processing platform, it supports statistics, reporting, querying and log analysis for input data, and responds / alerts on system problems. Through unsupervised machine learning on massive data, the abnormal log pattern is possible to be found automatically for potential system vulnerabilities and problems.
By connecting massive inputs from other systems, monitoring and alerting functionalities can be quickly provided in configurations. The stability and effectiveness of the system are fully tested in most of our use cases in our distributed platform. It can be applied as a module to our IoT solution, and works for manufacturing, energy, logistics and many other fields.
Overview of the Tech Stack
All of our applications are based on our own container and bigdata platform mentioned in the digram.
The container technology is the base of our Technologies.
Kubernetes helps us organizes and manages containers as a cluster, and adaptively scalable as a distributed platform. Some customizations are developed to meet the users' requirements of multi-tenancy, CI/CD, monitoring and other aspect.
DaaS layer accepts a lot of open-source bigdata components and develops some specific components for our applications' requirements.
The platform is adapted to physical clusters, AWS, Tencent Cloud and some other cloud solutions already, and is possible to be installed rapidly.
Stream Processing Framework
We build up our stream processing solution based on our tech stack, majorly on data processing for realtime and AI.
This model combines our clustering and big data capabilities, builds a reference of machine learning from modeling training to deployment, and provides an end-to-end solution on Bigdata and machine learning.
By reducing the effort to integrate AI algorithm and Bigdata streams, it provides a light-weight AI framework that can quickly precipitate AI assets into various industries for AI Ops.
The system integrates the realtime information reported by electric vehicles with the map data and displays it on the map in realtime. At the same time, it supports historical playback and positioning of specific vehicles. With the support of our data platform capabilities, we have effectively supported 30,000 vehicles per day for 100M events / day, the scalability and redundancy of our Bigdata platform is fully tested on the case.
In addition to car companies, the system can be quickly adapted to other map-related applications, such as taxi, delivery, outdoor sports and so on.
Based on unsupervised deep-learning model, interactive with experience people for their domain knowledge, the system can save a lot of effort on building new category system and automatically predict the right category based on descrptions. The result is connected to our reporting / monitoring system for system requirements.
The product can be used to other fields, for example banking, insurance, education, as well.
This is a container cloud platform based on kubernetes, and customized for user requirements.
Most of Bigdata components are integrated to the platform by using jsonnet, and can be installed by one command.
The platform can be installed on Physical cluster / AWS / AliCloud for in-cloud or on-premise requirements.
The annotation modeling system generates annotations based on the raw data in the context of the customer's business scenario. Based on this, it acts as a analysis tool for users to build their own rules and is possible to generate AI model automatically for clear target.
Lifecycle management of annotations is provided as well, for helping users quickly implement their own value.
3-year start-up / 10+ years engineering experience in Google / Intel
Experienced in high-performance and distributed computing
Xiaofeng lives in Fremont, CA email@example.com