Paras Gulati Email and Phone Number

- Researched and implemented various techniques and methods to develop a robust deep learning-based solution to identify and authenticate humans using their Electromygraphic (EMG) signatures. - Achieved an accuracy of 98% on 40 classes with an F1 score of 0.8.

· Designed and developed Change Data Capture (CDC) pipelines to incorporate real-time data ingestion from various data sources using the Kafka cluster and Apache MQ on AWS. · Designed an algorithm for graphification of tabular data by using statistical analysis on JanusGraph database. · Explored various unsupervised machine learning models to find anomalies in financial and healthcare data.

• Leading the AI research to develop strategic plans and road-maps to develop an end-to-end deep learning model that classifies EMG signals of various hand gestures for an advanced prosthetic arm.• Created an open-source python library to analyze, pre-process, visualize, and apply DSP filters to EMG samples.• Proposed a novel deep learning architecture based on Dilated CNN and LSTM that classifies 17 gestures with 86% accuracy, studied on 40 intact subjects.• Performed GradCAM analysis to highlight respective EMG signals that led to the decisions made by the model, thereby enhancing overall model transparency.

• Provided high-touch training to 10+ clients on data analysis using Python and developing machine learning models using Scikit-learn. • Designed curriculum in big data technologies (Hadoop, PySpark, Hive) and mentored clients to develop machine learning projects on distributed platforms.• Technology Stack: Sklearn, Pandas, Numpy, Matplotlib, Hadoop, PySpark, Hive, Pig

• Designed schema and developed applications for a broadband company toautomate workflows and add new functionalities to the billing system. Primarily utilized C++, Oracle, PL/SQL, and Unix for development.• Formulated workflows for Deutsche Bank to extract data from shared tables and load them to the target database after necessary transformations and validations. Used Informatica Power Center to perform the ETL operations.• End-to-end implementation of data pipelines to migrate data from backup tables to base tables using PL/SQL and bash scripting.• Accelerated the query execution time to extract bulk data from multiple tables by 30% using SQL optimization.• Provided production support to web applications of Deutsche Bank. Troubleshot and resolved production issues through the analysis of XML logs. Oracle, PL/SQL.

• Developed in depth understanding of FPGA and ASIC systems and used Xilinx software to simulate the model and create test-bench waveform• Used Verilog HDL to simulate Adders (Kogge Stone adder, Ripple Carry adder, etc), Subtractors, Multiplexers, Demultiplexers, Seven Segment Display, and Vedic Multiplier using different adders

• Developed a working prototype of a gesture controlled robotic car aimed at serving the physically challenged personnel using accelerometer, RF transmitter and receiver, and Atmega 16 microcontrollers. Used C++ to write the code.• Developed Line follower, Edge Avoiding robot, Digital Temperature Sensor, Traffic Light Prototype, Wireless Chat Protocol using AVR Microcontroller and C++ programming language• Received Training in embedded systems and robotics with a focus on AVR microcontrollers