Sooraj K. Email & Phone Number

Denver, Colorado, Us

New York, Ny, Us

New York, New York, Us

• Developed statistical and machine learning models to replace $100,000 worth of vendor-based advanced analytics over firm’s web proxy data• Utilized Amazon Sagemaker (AWS) to train and deploy models, Amazon Elastic MapReduce (AWS) to create a cluster of EC2 instances for the distributed training of models, and Amazon S3 (AWS) to store the training and testing datasets• Built and designed a data transformation platform that queries 7 days’ worth of raw web proxy data and then transforms the queried data into a numerical format so that it can be used to train the models developed in Amazon Sagemaker (AWS)• Enriched web proxy data with data representing a variety of malicious behavior using adversary simulation software• Implemented models using Apache Spark and Apache Hive via their Python libraries

A challenging problem in current hearing aid technology is source separation of a desired acoustic signal from an unwanted acoustic signal, such as speech, noise, and reverberations. Currently, the approach of blind-source separation, which does not utilize any information about the acoustic path between a speaker and a microphone, runs into difficulty when separating out the desired acoustic signal from the audio mixture when the room is reverberant.To circumvent this issue, the approach that we use is to aid the microphone array system in beamforming through the use of a pilot signal or so-called ‘beacon.’ The beacon is an audible-range or ultrasonic-range acoustic pilot signal that would be placed near the desired acoustic source. It would provide the needed acoustic path information - in the form of an acoustic impulse response - which would help a beamformer, along with the second-order statistics of the interference signals, to separate out the desired acoustic signal and increase its signal-to-noise ratio. Multiple lengths and frequency ranges of pilot signals are tested to determine the most useful pilot signal in a human environment.

New York, New York, Us

• Implemented data pipeline that ingests open-source cyber threat intelligence data from several open-source security and event management systems via their respective APIs.• Extracted more than 107,000 IPs, hashes, malicious hostnames and other indicators of compromise from the ingested open-source cyber threat intelligence data, used them to create STIX (industry-standard) objects and transferred the objects to Goldman’s internal threat-intelligence platform.• Utilized the retrieved open-source threat intelligence data to programmatically create over 1500 reports in Goldman’s internal threat intelligence platform• Deployed the data pipeline on the team’s production server and created fail-safe mechanisms such as automatic email generation and error logging to allow other security engineers to quickly locate and fix any issues

Dubai, Uae, Ae

• Wrote reports, which was published on Protiviti’s internal security blog, on how hacktivist groups threaten banks across the Middle East and how Protiviti could assist clients in protecting their IT systems. • Conducted research on the new VAT being introduced in several Middle Eastern nations and how Protiviti could help clients adapt their IT systems to the new VAT tax.

Master Of Engineering - Meng, Computer Engineering

Bachelor Of Science (B.S.), Computer Engineering

High School Diploma, General Studies