Citibank, N.a.

Citigroup Technology, Inc.

Citigroup, Inc.

Miriam Silver

Daniel Funken

Ashutosh K Sureka

Venkata Sesha Kiran Kumar Adimatyam

William Franklin Cameron

Manjit Rajaretnam

James Myers

Prithvi Narayana Rao

Pramod Goyal

Nimrod Barak

Prag Sharma

Avi Gefen

Sakthivel Palanivel

Stuart Murray

Payal Jain

Biraj Krushna Rath

Christopher James Byrne

Yogesh Abhyankar

Manojkumar Mahantappa Paragond

Steve Thomas Dayton

William George Cormier

Paola Gittino

Misty Stiers

Kirk M Lewis, Ii

Anurag Bhakoo

Vicky Kapadia

Toan Pham

Daniel Lewandowski

Vaibhav Kumar

Philip Davies

Ramkumar Ayyadurai

Ashutosh Pandey

Tariq Husayn Maonah

Mariusz Saternus

Chamindra Desilva

Vishal Mysore

Shardul Malviya

Samantha Cory

Deepak Jain

Wayne Liao

Girish Wali

Deepali Tuteja

Sofia Rahman

Prasanth Babu Madakasira Ramakrishna

Prashant Praveen

Ashim Mohanty

David Benamu

Jennifer Joy Bordon

Surendra Gangadhar Dumane

Mor Gelberg

Ashleigh Trout

Hila Gardi

Monique Huang

Hadar Karby

Saketh Ram Gurumurthi

Brett Pickard

Avi Levin

Systems and methods for constructing layered prompts to operate as input into a pre-trained large language model (LLM). The method involves obtaining a set of application domains in which the LLM will be used. Using these application domains, a set of guidelines is determined, defining operation boundaries for the LLM. A set of layers is determined, each associated with the guidelines and including variables representing attributes identified within those guidelines. Using these layers, a first layered prompt is constructed to test the initial operation boundaries of the guidelines and is supplied to the LLM to generate a set of responses. Based on the responses, a second layered prompt is dynamically constructed to test additional operation boundaries, ensuring iterative refinement and contextual relevance.

Layered multi-prompt engineering for pre-trained large language models

Presented herein are systems and methods for the employment of machine learning models for image processing. A mobile application for client-side image processing and validation, which interacts with and leverages native image processing software of the client device, where the image processing software and the mobile application include any number of machine-learning models for identifying a document and attributes of the document for recognition and validation. This mobile application uses the image processing software from a client operating system to control the camera. The image processing software generates various types of information about a video frame and the document, and the mobile application invokes APIs or software libraries of the image processing software to access the information and validate the frame and document.

Machine-learning models for image processing

Described herein are systems and methods for discovering and proactively mitigating previously unknown security vulnerabilities. The systems and methods herein can utilize security vulnerability information to discover potential security threats and can utilize this information to generate an attack using a machine learning model, such as a large language model. Generated attacks can be carried out to assess impact of a security vulnerability. An output can be provided that represents the assessed impact. In some implementations, the systems and methods herein generate patches or other mitigations for security vulnerabilities, which can be tested and deployed to address security vulnerabilities.

Generative cybersecurity exploit discovery and evaluation

Presented herein are systems and methods for the employment of machine learning models for image processing. A method may include a capture of a video feed including image data of a document at a client device. The client device can provide the video feed to another computing device. The method can include, by the client device or the other computing device object recognition for recognizing a type of document and capturing an image exceeding a quality threshold of the document amongst the frames within the video feed. The method may further include the execution of other image processing operations on the image data to improve the quality of the image or features extracted therefrom. The method may further include anti-fraud detection or scoring operations to determine an amount of risk associated with the image data.

Presented herein are systems and methods for the employment of machine learning models for image processing as may be performed by computing devices associated with an end user. A method may include obtaining video data comprising a plurality of frames including a document of a document type. The method may include executing an object recognition engine of a machine-learning architecture using image data of the plurality of frames, the object recognition engine trained to detect edges of documents. The method may include identifying, based on the edge detection, a plurality of boundaries for the document. The method may include validating, based on the plurality of boundaries, the document as the document type. The method may include transmitting via one or more networks, to a computer remote from the computing device, responsive to the validation of the type of document, the image data for the plurality of frames depicting the document.

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.

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

Systems and methods for novel approaches and/or improvements to real-time data processing of unstructured data. In particular, the systems and methods describe real-time data processing of unstructured data without interstitial standardization. For example, the systems and methods describe real-time data processing of unstructured data in which both the input and the output to the data processing pipeline is unstructured data.

Systems and methods for real-time data processing of unstructured data

Presented herein are systems and methods for regularly updating computer-form files. A method may include obtaining, by a computer, raw data containing a plurality of data records associated with a customer from a plurality of databases, in response to detecting an error in an error data record, automatically correcting executing a machine learning architecture, the error. The method may include, for each data record, determining a data category indicating one or more computer-form files for a data entry of the data record based upon a preconfigured mapping between a type of data of the data entry mapped to the data category, and in response to detecting a new customer data requirement, updating the one or more computer-form files associated with each data category according to each data record of each daily interval, and at a preconfigured time, generating the one or more computer-form files.

System and method for regular updates to computer-form files

A system or technique can provide a graphical user interface (GUI) with input elements, each associated with different attributes of an individual to be depicted in a computer-implemented presentation. The system receives input data based on user interaction with a GUI input element and queries a database containing image metadata linked to a set of images. These images are collected based on metrics from prior presentations, and the metadata includes classification information provided by the individuals depicted. The system locates data in the metadata corresponding to the user input, identifies additional relevant data, and renders a second GUI input element based on this data. It then locates and retrieves an image from the set of images using the metadata, displaying the image in the GUI.

Graphical user interface to adjust image selection based on classification information

Systems and methods for streamlining model risk documentation platform outputs using natively sourced kernels

The disclosed data generation platform enables generation of an output in response to an output generation request based on tuning a routing model that enables model selection in a dynamic, system-sensitive manner. For example, the disclosed data generation platform receives an output generation request for a user device and generates a risk indicator associated with the output generation request. The platform can determine a current system state and generate a set of performance indicators and associated weighting values based on the risk indicator and the system state. The data generation platform can select a first routing model based on the weighting values. The data generation platform can provide the output generation request to the first routing model to generate an indication of a model with which to generate a model output responsive to the input. The data generation platform can enable access to the generated model output.

Latency-, accuracy-, and privacy-sensitive tuning of artificial intelligence model selection parameters and systems and methods of the same

Growing community of inventors

Tel Aviv, Israel

Miriam Silver