CV

Education

• Ph.D. in Electrical and Computer Engineering, Binghamton University (SUNY), Vestal, NY
  Expected May 2028
  Ph.D. studies transferred from Michigan Technological University, Computer Engineering (Aug. 2024 – Aug. 2026).

• M.S. in Data Science, Michigan Technological University, Houghton, MI
  Aug. 2022 – Apr. 2024
  GPA: 3.75

• B.E. in Information Science and Engineering, New Horizon College of Engineering, Bengaluru, India
  Aug. 2018 – May 2022
  GPA: 3.52

Research Experience

Robotics, Locomotion and Applied Control (RoLAC) Lab

Graduate Research Assistant
Binghamton, NY / Houghton, MI
May 2023 – Present

Multi-Fault-Tolerant Humanoid Locomotion – Unitree G1

• Developed SAC-based reinforcement-learning controllers for goal-directed locomotion of a 23-DoF Unitree G1 using Python, MuJoCo, Gymnasium, and Stable-Baselines3.
• Increased simulated course completion under two joint-lock faults from 0% to 72%.
• Implemented damage-propagation reduction using per-joint torque limits, reducing representative compensatory knee peak torque from 70.9 to 52.6 N·m under a hip-joint lock.
• Trained fault-tolerant crawling policies, increasing survival with forward progress from 65% to 92.5% under three simultaneous knee/ankle joint locks with the hip joints available in simulation.

Energy-Regularized Bipedal Locomotion under Varying Gravity – Bolt

• Funded by the National Science Foundation (NSF).
• Designed and trained PPO-based locomotion policies in MuJoCo across 26 energy-regularization settings under Earth and lunar gravity without prescribing target velocity, gait symmetry, or gait structure.
• Quantified emergent gait transitions using contact ratio, ground reaction forces, friction utilization, speed, and cost of transport.
• Identified gradual running-to-skipping transitions under Earth gravity and sharper hopping-to-loping transitions under lunar gravity.

RL-Based Bipedal Locomotion – Cassie

• Designed and trained PPO-based locomotion policies in MuJoCo for dynamic bipedal locomotion.
• Reduced joint-tracking error by 16%.
• Developed neural predictors for joint-trajectory estimation and investigated learning-based optimization of locomotion performance.

Reinforcement Learning for Humanoid Perception, Planning, and Control

• Led a systematic review of reinforcement learning for humanoid robots spanning 2000–2025.
• Organized methods across perception, planning, and control.
• Consolidated an 870-paper database from Scopus and IEEE Xplore, retaining 168 studies after systematic screening.
• Analyzed 45 representative works across algorithms, simulation environments, physical humanoid platforms, applications, and real-world deployment.

Scientific Machine Learning – FractionalNet

• Partially funded by Oak Ridge Associated Universities (ORAU).
• Developed a symmetric neural-network architecture for approximating half-order fractional derivatives from integer-order data.
• Designed a two-stage genetic algorithm for network hyperparameter optimization.
• Evaluated seven weight-initialization strategies and identified He-Uniform as the strongest-performing initialization for the three-layer architecture.

Industry Experience

Aeronautical Development Establishment (ADE), DRDO

Software Engineer Intern
Bengaluru, India
Feb. 2021 – Apr. 2021

• Developed software for UAV control-data processing and automated reporting.
• Improved analysis efficiency by 40% and reporting speed by 30%.

Teaching Experience

EE 3010 – Circuits and Instrumentation for CPS

EE 3010 – Circuits and Instrumentation for CPS

EE 3171 – Microcontroller Applications

EE 2174 – Digital Logic Design

Engineering Statistics and Research Mentoring

Skills

• Programming & Reinforcement Learning: Python, PyTorch, Stable-Baselines3, Gymnasium, RLlib, Scikit-learn, MATLAB, C++
• Robotics Simulation & Hardware Platforms: MuJoCo, Isaac Lab, ROS, Unitree G1, Bolt
• Embedded & Systems: Linux, Git/GitHub, Verilog, ARM Microcontrollers, Intel RealSense, Jetson Nano

Publications

FractionalNet: a symmetric neural network to compute fractional-order derivatives

D. Biligiri, J. A. Smitterberg, S. Akki, B. Goodwine, and T. Chen. (2025). "FractionalNet: a symmetric neural network to compute fractional-order derivatives" Nonlinear Dynamics, 113, 33381–33398.

Misc Projects

Multi-Ingredient Food Image Classification & Recipe Recommendation

• Built an end-to-end computer vision and NLP pipeline using YOLOv8n for multi-ingredient detection.
• Achieved mAP@50 = 0.829.
• Used TF-IDF retrieval and T5-based recipe generation.
• Deployed an interactive Gradio interface for image-based recipe recommendation.
• Technologies: Python, YOLOv8, OpenCV, Optuna, TF-IDF, Hugging Face T5, Gradio

Underwater Object Detection with Image Enhancement

• Evaluated YOLOv8 object detection on original and Semi-UiR-enhanced underwater imagery.
• Image enhancement improved visual clarity by 12%.
• Detection on original imagery achieved 30% higher accuracy than detection on enhanced images.
• Technologies: Python, OpenCV, YOLOv8, Ultralytics, Semi-UiR

Honors & Awards

• Jonathan Bara Outstanding Graduate Teaching Assistant Award, Michigan Technological University, 2025
• GSG Outstanding Graduate Teaching Assistant Award, Michigan Technological University, 2026

Service & Leadership

• Vice President, Graduate Student Government, Michigan Technological University
  Apr. 2026 – Aug. 2026

• Department Representative, Graduate Student Government, Michigan Technological University
  Aug. 2023 – Apr. 2026

• Student Member, IEEE

• Tutor, McNair Scholars Program, Michigan Technological University
  Aug. 2023 – Apr. 2024