Dhruv Nandakumar
Boston,United States
Summary
Accomplished Senior AI Engineer at NVIDIA with expertise in deep learning and software design patterns. Spearheaded projects that enhanced agent accuracy by 200% and reduced inference latency by 35%. Proven leadership in managing teams and delivering high-impact AI solutions, driving innovation in complex environments.
Overview
5
5
years of professional experience
Work History
Senior AI Engineer
NVIDIA
Boston, MA
04.2024 - Current
- Led a project on the NVIDIA Agent Toolkit Profiler and Optimizer, achieving up to a 200% improvement in agent accuracy autonomously on standard benchmarks.
- Led the design and development of novel Inference Time Scaling modules for the NVIDIA Agent Toolkit, unlocking up to 45% accuracy improvements in reasoning models, at 33% lower cost on industry-standard reasoning benchmarks.
- Led implementation of GPU-level distributed inference optimization strategies for agents, resulting in 35% decrease in LLM inference latency at data center scale by utilizing future workload-aware request routing and eager token prefill algorithms.
- Engineered GPU-native video preprocessing algorithms enhancing throughput of video processing for Vision-Language-Action model (Eagle 2) pre-training, outperforming traditional pipelines by 6x.
Lead Data Scientist (Applied Research)
Deloitte & Touche LLP
Boston, United States
09.2020 - 04.2024
- Managed and led six AI model development work streams, overseeing 10 data scientists with on-time delivery and zero cost overruns.
- Spearheaded advanced deep learning methods for cyber threat detection, achieving a 70% increase in end-user productivity.
- Developed methodologies for Large Language Models, Graph Neural Networks, and Deep Reinforcement Learning, resulting in models exceeding 99% precision and recall.
- Authored over 10 peer-reviewed manuscripts presented at international IEEE conferences.
- Implemented inference methods for real-time model deployment, achieving a 40x performance boost on over 10 million events per second.
- Utilized NVIDIA Morpheus and Neo4j for parallelized inference in cloud environments, enhancing processing efficiency.
Education
Master of Science in Engineering - Data Science
University of Pennsylvania
Pittsburgh, PA06-2026
Bachelor of Science - Computer Science
University of California, Davis
Davis, CA01.2020
Books and Patents
- Book: Reinforcement Learning for Cyber Operations: Applications of Artificial Intelligence for Penetration Testing (2024)
- Patent: Task Agnostic Planning and Execution of Agent Workflows (Filed 2025)
- Patent: Online Inference Optimization and Control of Agent Workflows (Filed 2025)
- Patent: Generalized Human Feedback Collection for Running Agent Workflows (Filed 2025)
- Patent: Offline Inference Performance Analysis for Agentic Workflows (Filed 2025)
Research Publications
- Soumya Banerjee, Sandip Roy, Sayyed Farid Ahamed, Devin Quinn, Marc Vucovich, Dhruv Nandakumar, Kevin Choi, Abdul Rahman, Edward Bowen, Sachin Shetty, MIA-BAD: An Approach for Enhancing Membership Inference Attack and its Mitigation with Federated Learning., ArXiv abs/2312.00051, 2023
- Sathvik Murli, Dhruv Nandakumar, Prabhat Kushwaha, Cheng Wang, Christopher Redino, Abdul Rahman, Shalini Israni, Tarun Singh, Edward Bowen, Cross-temporal detection of novel ransomware campaigns: A multi-modal alert approach., [2309.00700], 09/01/23, http://export.arxiv.org/abs/2309.00700
- Dhruv Nandakumar, Devin Quinn, Elijah Soba, Eunyoung Kim, Christopher Redino, Chris Chan, Kevin Choi, Abdul Rahman, Edward Bowen, Foundational Models for Malware Embeddings Using Spatio-Temporal Parallel Convolutional Networks., Pre-print under review, https://arxiv.org/abs/2305.15488
- Dhruv Nandakumar, Sathvik Murli, Ankur Khosla, Kevin Choi, Abdul Rahman, Drew Walsh, Scott Riede, Eric Dull, Edward Bowen, A Novel Approach to User Agent String Parsing for Vulnerability Analysis Using Multi-Headed Attention., 2023 International Conference on Machine Learning and Cybernetics (ICMLC 2023)
- Cheng Wang, Christopher Redino, Abdul Rahman, Ryan Clark, Daniel Radke, Tyle Cody, Dhruv Nandakumar, Edward Bowen, Discovering Command and Control (C2) Channels on Tor and Public Networks Using Reinforcement Learning., Pre-print under review
- Riddam Rishu, Akshay Kakkar, Cheng Wang, Abdul Rahman, Christopher Redino, Dhruv Nandakumar, Tyler Cody, Ryan Clark, Daniel Radke, Edward Bowen, Enhancing Exfiltration Path Analysis Using Reinforcement Learning., Pre-print under review
- Dhruv Nandakumar, Robert Schiller, Christopher Redino, Kevin Choi, Abdul Rahman, Edward Bowen, Matthew Weeks, Aaron Shaha, Joe Nehila, Zero Day Threat Detection Using Metric Learning Autoencoders., 21st International Conference on Machine Learning Applications (ICMLA-22), 2022
- Christopher Redino, Dhruv Nandakumar, Robert Schiller, Kevin Choi, Abdul Rahman, Edward Bowen, Matthew Weeks, Aaron Shaha, Joe Nehila, Zero Day Threat Detection Using Graph and Flow Based Security Telemetry., International Conference on Computing, Communication, and Intelligent Systems., 2205.02298, https://arxiv.org/abs/2205.02298, 2022
- Deepak Kushwaha, Dhruv Nandakumar, Akshay Kakkar, Sanvi Gupta, Kevin Choi, Christopher Redino, Abdul Rahman, Sabthagiri Saravanan Chandramohan, Edward Bowen, Matthew Weeks, Aaron Shaha, Joe Nehila, Lateral Movement Detection Using User Behavioral Analysis., Submitted 37th AAAI Conference on Artificial Intelligence., 2208.13524, https://arxiv.org/abs/2208.13524, 2022
- Marc Vucovich, Amogh Tarcar, Penjo Rebelo, Narendra Gade, Ruchi Porwal, Abdul Rahman, Christopher Redino, Kevin Choi, Dhruv Nandakumar, Robert Schiller, Edward Bowen, Alex West, Sanmitra Bhattacharya, Balaji Veeramani, Anomaly Detection via Federated Learning., 21st International Conference on Machine Learning Applications (ICMLA-22), 2022
Timeline
Senior AI Engineer
NVIDIA
04.2024 - Current
Lead Data Scientist (Applied Research)
Deloitte & Touche LLP
09.2020 - 04.2024
Master of Science in Engineering - Data Science
University of Pennsylvania
Bachelor of Science - Computer Science
University of California, Davis