Make AI Practically

Research Areas

Natural Language Processing
Natural Language Processing

To enable computer systems to interact with the outside world in the form of natural language text, track and research the most advanced natural language text understanding technology, and incubate more usable natural language processing technology and business scenarios. Research fields: Text classification, information extraction, knowledge mapping, etc.

Machine Learning
Machine Learning

Have a computer execute automatically without explicit programming. Research fields: Machine learning theory, numerical optimisation, heterogeneous computing; Supervision, no supervision.

Computer Vision
Computer Vision

Computer vision allows computers to understand the real visual world without explicit programming. Research fields: Graphic analysis, object recognition, text recognition.

Applicable Fields

Intelligent O&M
Intelligent O&M

Core areas: finance, games; It is applied to banking and game business with SaaS support for quick use and immediate output.

Intelligent O&M
Financial Marketing
Financial Marketing

Core area: Finance; Applied to securities and banking business, it emphasises interpretability and empowerment to help better long-term development.

Financial Marketing
Target detection
Target detection

Core areas: government affairs; Used in a subway and medical treatment, high business identification, better landing effect.

Target detection

Application Scenarios

AI+ Banking Case -- Customer churn Solution
Customer Pain Points

The retention cost is too high and the effect is weak.

How to deal with loss in advance and reduce retention costs?

Automatic iteration

Automatic data management and model update, multi-dimensional cost reduction.

One-button start

One-button start, visualise output process and results.

White Box

White-box process, on-site technology empowerment, providing sustainable solutions.

More powerful

Predict further, about ¥6 million RMB in quantifiable gains.

AI+ Government affairs case -- Intelligent guidance solution in Subway
Customer Pain Points

Fare evasion is easy to occurs at crowded stations. In 2018, Shanghai Metro collected about 10.35 million yuan in compensation fares, and the loss caused by fare evasion cannot be ignored. However, the station staff's irregular supervision of fare evasion will result in a massive workforce waste, with an average of 0.2 people being spent in each station supervising fare evasion. 


Realise automatic work by making devices more intelligent.

Expand convenience

Reduce the construction complexity,  and make expansions more convenient.

Passage Time

Guide light passage time.

Convenient installation

Make the installation of the lightbox convenient and fast.

AI+ Medical -- Cell recognition

High accuracy

The accuracy is up to 83%.


Combined with BI large screen visualisation.

Expert post-processing algorithm

Combine the experience of dozens of medical experts on social networks as the last line of defence for results. To do extraction for characteristic computing.

AI+ RESUME Analysis - Smart resume analysis

LDA parsing


Parsing each module of a resume and identifying the topic that each module belongs to, such as education and work experience.

Hidden Markov lexical analysis


Extract the key contents of each module, such as educational background, work experience, etc.

Relational reasoning database


Support relational reasoning. For example, if the candidate graduated from a school with a good ranking in mathematics, it can be inferred that the candidate is more suitable.

Explore Learning

Our primary research areas are unsupervised learning, deep learning, graph learning, mainly applied in intelligent operation and maintenance, fintech and other fields. Based on our science and technology development strategy of PES (Professional service, Enabling ecology, Science and technology for good), we create instant, efficient, simple and easy to use AI technology, and explore innovation in the field of intelligent operation and financial technology. At present, we have created root cause positioning, abnormal detection, jitter noise reduction and other technologies for 'Weiying Cloud' (intelligent operation and maintenance) products.


The team consists of senior algorithm engineers from Innovation Business Department, Delivery and Service Department, Product Development Department, and Public Cloud Technology Operation Service Department.