Systems and methods for searching and storing data over a computer network

Patent metadata

Field	Value
Publication number	US-12450243-B1
Application number	US-202519204415-A
Country	US
Kind code	B1
Filing date	May 9, 2025
Priority date	Dec 18, 2023
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

Systems and methods for streamlining risk modeling in software development using natively sourced kernels are described. The system may receive a native kernel for the first model, wherein the native kernel comprises a native code sample and a native description of the native code sample. The system may input the native code sample into an artificial intelligence model to generate a first output. The system may filter the first output based on the native description to generate a first validation assessment for the first model. The system may generate for display, in the user interface, the first validation assessment.

First claim

What is claimed is:

1. A system for searching and storing data over a secured computer network, the system comprising:
one or more hardware processors; and
one or more computer-readable media comprising instructions recorded thereon that when executed by the one or more hardware processors cause operations comprising:
generating an encrypted datastore by:
receiving a plurality of secure applications;
parsing the plurality of secure applications determine a plurality of labeled code samples;
parsing the plurality of labeled code samples to determine labeled code sample characteristics corresponding to one or more of the plurality of labeled code samples; and
encrypting the plurality of labeled code samples to generate a plurality of encrypted labeled code samples;

receiving, via a user interface, a first user input comprising a first search request for code samples in the encrypted datastore that have a first characteristic;
generating a first encrypted feature input based on the first search request;
inputting the first encrypted feature input into a first trained model to receive a first encrypted output, wherein the first trained model compares code sample characteristics described by inputted first features inputs to the labeled code sample characteristics corresponding to one or more of the plurality of encrypted labeled code samples to generate outputs identifying one or more of the plurality of encrypted labeled code samples;
generating a second encrypted feature input based on the first encrypted output;
inputting the first encrypted output into a second trained model to receive a second encrypted output, wherein the second trained model generates human-readable descriptions related to the labeled code sample characteristics of the one or more of the plurality of labeled code samples identified by feature based on inputted second feature inputs; and
generating for display, in the user interface, a first recommendation of results corresponding to the first search request based on the second encrypted output, wherein the first recommendation is unencrypted.

2. A method for searching and storing data over a computer network, the method comprising:
accessing a datastore populated by:
receiving a plurality of applications;
parsing the plurality of applications determine a plurality of labeled code samples; and
parsing the plurality of labeled code samples to determine labeled code sample characteristics corresponding to one or more of the plurality of labeled code samples;

receiving, via a user interface, a first user input comprising a first search request for code samples in the datastore having that have a first characteristic;
generating a first feature input based on the first search request;
inputting the first feature input into a first trained model to receive a first output, wherein the first output comprises a labeled code sample comprising a first labeled characteristic that corresponds to the first characteristic, and wherein the first trained model compares code sample characteristics described by inputted first features inputs to the labeled code sample characteristics corresponding to one or more of h plurality of labeled code samples to generate outputs identifying one or more of the plurality of labeled code samples;
generating a second feature input based on the first output;
inputting the first output into a second trained model to receive a second output, wherein the second output comprises a human-readable description of the labeled code sample, and wherein the second trained model generates descriptions related to the labeled code sample characteristics of the one or more of the plurality of labeled code samples identified by feature based on inputted second feature inputs; and
generating for display, in the user interface, a first recommendation of results corresponding to the first search request based on the second output.

3. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving native code sample for an application under development;
determining that the native code sample comprises the first characteristic; and
populating the first search request with the first characteristic based on determining that the native code sample comprises the first characteristic.

4. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving native code for an application under development;
determining a first code sample boundary in the native code;
parsing a portion of the native code based on the first code sample boundary; and
determining the first characteristic based on the portion.

5. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving native code for an application under development;
determining a first containerized portion in the native code; and
parsing the first containerized portion to determine the first characteristic.

6. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving a native code sample for an application under development;
determining a code dependency in the native code sample; and
determining the first characteristic based on the code dependency.

7. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving native code for an application under development;
performing a runtime analysis of the native code in a sandbox environment; and
determining the first characteristic based on the runtime analysis.

8. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving a native code sample for an application under development;
tokenizing the native code sample to generate a token representation;
determining a syntax tree based on the token representation; and
determining the first characteristic based on the syntax tree.

9. The method of claim 2 , wherein the receiving of the first user input further comprises:
receiving a native code sample for an application under development;
determining a naming convention in the native code sample; and
determining the first characteristic based on the naming convention.

10. The method of claim 2 , wherein the inputting of the first feature input into the first trained model to receive the first output further comprises:
determining a first code function corresponding to the first characteristic; and
comparing the first code function to a labeled code function corresponding to the first labeled characteristic.

11. The method of claim 2 , wherein the inputting of the first feature input into the first trained model to receive the first output further comprises:
determining a data value corresponding to the first characteristic; and
comparing the data value to a plurality of data values in the labeled code sample.

12. The method of claim 2 , wherein the inputting of the first feature input into the first trained model to receive the first output further comprises:
determining a similarity between the first characteristic and the first labeled characteristic; and
comparing the similarity to a threshold similarity.

13. The method of claim 2 , wherein the inputting of the first feature input into the first trained model to receive the first output further comprises:
determining a first data source for a native code sample corresponding to the first characteristic; and
comparing the first data source to a labeled data source corresponding to the first labeled characteristic.

14. The method of claim 2 , wherein the inputting of the first

Assignees

• Citibank Na

Inventors

• Silver Miriam
• Barak Nimrod
• Sharma Prag
• Gefen Avi

Classifications

• G06F21/577

  Assessing vulnerabilities and evaluating computer system security · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F16/2458 Primary

  Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries · CPC title

• G06F16/2428

  Query predicate definition using graphical user interfaces, including menus and forms (G06F16/2423 takes precedence) · CPC title

• G06F8/75 Primary

  Structural analysis for program understanding · CPC title