Kumar Arnav Email & Phone Number

Calabasas, California, United States

Calabasas, California, United States

• Developed an advanced voice-to-voice bot model hosted on AWS, enabling multilingual interactions by processing user inputs and generating responses in the same language.
• Implemented novel interruption and user speech understandability features:
• Enabled the LLM to determine when a user had finished speaking, allowing it to wait for the user to complete their thoughts.
• Accurately detected user interruptions, enhancing the fluidity and responsiveness of interactions.
• Integrated backend systems and ensured model feasibility to support optimal performance.
• Trained the model, focusing on improving accuracy and responsiveness to user inputs.

Los Angeles, California, United States

• Constructed a reinforcement machine learning-based model for Incremental View Maintenance(IVM) techniques for PostgreSQL achieving approx 53% faster runtime for all queries.
• Explored the effectiveness of using reversed input files in singing voice deepfake detection, demonstrating significant improvements in Equal Error Rate (EER) across various fusion methods.
• Conducted and completed experiments using RawNet and LFCC models, evaluating their performance in detecting deepfake singing voices.
• Collaborating with a research team to finalize and present detailed EER comparisons, contributing to advancements in the field of deepfake detection.

Remote

• Deployed a Random Forest model to surmise customer churn for an e-commerce company, resulting in a 18% reduction in churn rate
• Trained a Gradient Boosting Machine (GBM) algorithm to optimize advertising campaigns for a marketing agency, increasing click-through rates by 27%
• Trained a Support Vector Machine (SVM) model to anticipate machine failure for a manufacturing company, reducing unplanned downtime by 32%
• Performed Natural Language Processing (NLP) techniques deploying LSTM networks for sentiment analysis for a social media analytics firm, boosting brand monitoring accuracy by 78%

• Applied vision system for an autonomous drone deploying Convolutional Neural Networks (CNN) to identify and differentiate objects and obstacles, enhancing navigational accuracy by 59%
• Accomplished object recognition and distance estimation algorithms, improving drone's retrieval success rate by 80%

Work From Home

* Led development of a user-centric Android app (Java) for matching students and teachers based on availability of both students and teachers, enhancing user convenience and contributing to a 4.5-star rating on the Google Play Store* Directed backend team integrating Firebase for efficient data management, leading to a 82% improvement in data.

Lucknow Area, India

* Identified vulnerabilities and breakpoints in 300+ websites and Android applications for clients as an ethical hacker, operating under strict non-disclosure agreements prevent sharing of specific details* Performed comprehensive security audits for mid-sized companies, resulting in a 70% improvement in cybersecurity practices based on confidential.

Master Of Science - Ms, Artificial Intelligence

Activities and Societies: Boxing, stock tradingCourses: 1. Analysis of Algorithm 2. Computer Science Machine Learning 3. Applied Natural.

Bachelor Of Technology - Btech, Computer Science

Morgan Stanley Investment Banking 101