Legal Processes Such as Analyzing

[sadiksojib35]

On the topic: Machines won't rise up, but you can get fired from your job: we figure out why you need to master neural networks Transition to cascade systems With the deeper penetration of machine learning into our company's business processes, with the expansion of the context and the complexity of logic, there was a need to transform the decision-making system, and simple linear algorithms were replaced by cascade systems.

Over time, we began to notice duplicate blocks in our bahamas telegram database services. To manage them effectively, we began to combine them into separate product (features store, credit robot, collection and judicial robots, LTV, offering) and technical (monitoring and logging, CI\CD + MLOps) logical services. We also took into account new market requirements when changing the structure.

If earlier we could wait for a decision to be made for several minutes, sometimes up to half an hour, now this time has been reduced. The decision must be made within a few seconds, otherwise the conversion begins to fall sharply. On the topic: The next stage of neural network development: what is interactive AI and why is it “smarter” than generative ML Team Instead of a Lonely Data Scientist The more complex the system becomes, the more tasks appear, the deeper each of the areas of processing and solution preparation.

One specialist can no longer cope, and then the team grows in accordance with the needs of the business. At the very beginning, there was only one role in the R&D department of our company - Data scientist, who prepared his own data, did research related to the search for optimal models, trained models, carried out performance of various indicators and presented reports.

The team looks completely different now. Who's on it: data engineer, who deals with data preparation, ETL, collecting data from sources and converting it into machine-readable form; feature engineer, who comes up with ways to transform data into features that describe a real object or subject (in our case, the borrower); software engineer, who ensures the implementation of services based on prepared models; MLOps engineer is a new and interesting role, similar to DevOps engineer.

This specialist works in the field of Data science, provides automation of pipelines, training and retraining of models, monitoring, implementation of a number of automatic calculations and delivery of machine learning services to the final production state. New structure of the service The introduction of ML services into the activities of microfinance companies could not but affect all subsequent business processes.