Shaifalee Saxena

Shaifalee Saxena

Ph.D. student at UNM and Graduate Research Assistant at Los Alamos National Laboratory working on reinforcement learning, robotics, and control.

CV

Download CV PDF

Education

Research Experience

Graduate Research Assistant, Los Alamos National Laboratory

Adaptive Machine Learning Group (AOT-IC) | 2025-present
Working on reinforcement learning for accelerator tuning and control under the supervision of Dr. Alexander Scheinker.

Research Assistant, University of New Mexico

MARHES and AgMan Labs | 2024-present
Working on hybrid control methods for space robotics under the supervision of Dr. Rafael Fierro.

Current Research Projects

Robust Deep RL for Time-Varying Control

Developing hybrid learning-control methods that combine deterministic actor-critic reinforcement learning with bounded extremum seeking for robustness under drift and nonstationarity.

Multi-Task Reinforcement Learning for Robotic Manipulation

Training a single policy to perform multiple manipulation tasks such as push, grasp, and pick-and-place with shared representations and task-conditioned behavior.

Vision Representation Learning for RL via Autoencoders

Building latent-space control pipelines that compress image observations into compact latent states for downstream reinforcement learning.

Space Robotics: Solar Panel Deployment Failure Detection and Repair

Developing vision-based failure detection and corrective control strategies for stuck or jammed deployment mechanisms.

Publications

Invited Talks

Professional Experience

Airbus, India

Senior Flight Controls Engineer | 2021–2024

Tata Aerospace and Defense, India

Senior Control Systems Engineer | 2018–2021

Skills

Controls and Modeling

System identification, nonlinear dynamics and modeling, discrete-time and continuous-time systems, frequency-domain analysis, filtering, PID and loop-shaping, state-space control, state estimation, observers, and simulation workflows.

Reinforcement Learning and Machine Learning

Deep reinforcement learning, actor-critic methods, DDPG, hybrid learning-control methods, representation learning, autoencoders, multi-task learning, and data-driven policy evaluation.

Robotics and Simulation Software

ROS 2, Gazebo, MuJoCo, MATLAB/Simulink, SCADE, ADAMS, and ANSYS.

Hardware and Sensors

UR5e manipulator, quadcopters, RGB cameras, gimbals, Arduino, Raspberry Pi, ultrasonic sensors, and LiDAR.

Honors and Awards