Siri Gadipudi

I am a graduate student pursuing Master of Science in Electrical and Computer Engineering, at the University of Washington, Seattle.

My primary concentration is in the field of Machine Learning and Robotics. I am currently a Graduate Student Researcher at the WEIRD lab in Paul G Allen School of Computer Engineering, with my Advisor: Prof. Abhishek Gupta. My research work currently is in the field of Deep Reinforcement Learning and Autonomous Robot Learning.

Prior to my Master's, I served as a Scientist at the Indian Space Research Organization (ISRO), where my work was focused on AI-driven solutions for Radars and Tracking. My expertise in the role extended to real time operations and system design where I managed console operations for various ISRO missions and also executed different technology development projects like the development of the data processing software for the radar console and display.

In 2020, I graduated with a Bachelor of Technology degree in Electronics and Communications Engineering (with a specialization in Avionics) from the Indian Institute of Space Science and Technology.

Open to Work

I’m on track to graduate in June 2024 and I’m actively seeking full time job opportunities, specifically roles such as Machine Learning Engineer, Applied Scientist, AI Engineer, Data Scientist, or Robotics Engineer. I am also open to internship positions in Winter/Spring 2024.

Email  /  Resume  /   /  Github  /  LinkedIn

profile photo
News
  • 12/2023: Attended and presented our poster (RePo) at the NeurIPS 2023 conference at New Orleans, LA. Also attened the Women in ML (WiML) workshop at the conference.
  • 10/2023: Attended the Women Engineer's 2023 conference by Society of Women Engineers at LA
  • 10/2023: Our paper "RePo: Resilient Model Based Reinforcement Learning by Regularizing Posterior Predictability" has been accepted to NeurIPS 2023 as a Spotlight paper.
Publications
  • Zhu, C., Simchowitz, M., Gadipudi, S., & Gupta, A., (NeurIPS 2023 Spotlight). RePo: Resilient Model-Based Reinforcement Learning by Regularizing Posterior Predictability. Read the paper here: arXiv:2309.00082
  • S. Gadipudi, P. P. Rajeevan. and R. S. Kaarthik, ’A Grid Connected Open-end Winding Induction Generator System with Series Compensation’ published in Early Access area on IEEE Xplore and is accepted for publication further in the IEEE Transactions on Industry Applications, 2021, DOI: 10.1109/TIA.2021.3128377
  • S. Gadipudi, P. P. Rajeevan. and R. S. Kaarthik, ”A Grid Connected Open-end Winding Induction Generator System with Series Compensation”, 2020 IEEE International Conference on Power Electronics, Smart Grid and Renewable Energy (PESGRE2020), 2020, pp. 1-4, DOI: https://ieeexplore.ieee.org/document/9070498
Professional Experience

    Radar Systems Engineer/Scientist - SC, Indian Space Research Organization, Sriharikota, India; January 2021 - June 2022:

  • Optimized Radar Data Processing: Implemented advanced ML algorithms (CNNs and RNNs) resulting in a 20% increase in detection accuracy and a 15% improvement in tracking precision for radar imagery, significantly enhancing situational awareness and bolstering radar system capabilities.
  • AI-Enhanced Radar Operations: Successfully integrated AI-driven solutions into radar operations, leading to a 30% reduction in noise levels and 95% accurate target classification. This real-time analysis of radar data significantly improved system reliability and enabled more accurate and efficient object tracking in complex environments.
  • Real-time Console Operations: Proficiently managed real-time console operations for various ISRO missions namely RH200, PSLV-C51, RH560, GSLV-F10, and PSLV-C52, ensuring seamless radar functionality.
  • Data Processing System Software Development: Developed the data processing system software in C++ for efficient data transmission and collection between different subsystems of the Radar. Designed the console's GUI in Qt platform and integrated all subsystems' hardware through FPGA.
  • System Design: Designed and implemented angle control algorithms for the Azimuth and Elevation angle tracking subsystems of the Radar, contributing to improved tracking precision and operational efficiency.
  • Maintenance Activities: Incharge of maintenance activities for the angle tracking system related drives and its related control systems, receiver sub-system, two transmitters that can transmit one mega Watt power in RF frequency, digital systems, high power devices, antenna servo motors, and the data processing software system.
Research Experience

    Machine Learning/ Deep Robotic Learning Researcher, Washington Embodied Intelligence and Robotics Development (WEIRD) Lab

    Paul G. Allen School of Engineering, University of Washington, Seattle; March 2023 - Present:

    Advisor: Abhishek Gupta

    Project: Human-Guided Interventions for Reset-free Autonomous Reinforcement Learning

  • Designing a novel methodology to introduce human-guided interventions for reset-free RL algorithms, enabling efficient autonomous robotic learning.
  • Achieved better performance and ease of implementing deep RL algorithms for visual observations by developing a tracker module using object segmentation models (SAM and FC-CLIP) customized for object tracking.
  • Other projects and their content coming up soon.

    • Undergraduate Student Researcher, Department of Avionics

    Indian Institute of Space Science and Technology; June 2019 - July 2020

    Advisors: Rajeevan P P and Sudharshan Kaarthik

    Project: Series Reactive Power Compensation for Grid Connected Open-end Winding Induction Generation Systems for Wind Energy Applications

  • A new series reactive power compensation scheme was proposed for a grid connected induction generator system where it utilises both ends of the induction machine with one end of the winding connected to the grid and the other end connected to a capacitor fed voltage source inverter which supplies the required reactive power for the excitation of the induction generator.
  • Performed extensive research, simulations, analysed the new proposed concept and designed a new current oriented control scheme where active power is being delivered to the grid by the induction generator without drawing reactive power from it.
  • A grid connected open-end winding induction generator system with series reactive power compensation was first simulated in PLECS 4.3.1 software and then implemented on hardware for experimental validation using a digital signal processor (DSP TMS320F28335).
  • A current oriented rotating reference frame based closed loop control scheme was implemented on hardware so that active power gets delivered to the grid from the open-end winding induction generator without drawing any reactive power from the grid for its excitation.
  • The reactive power requirement for the excitation of induction generator is compensated by a capacitor-fed voltage source inverter so that it reduces the burden on the grid to supply reactive power.
  • The proposed model was simulated and studied for the dynamic case where the speed of the induction machine is transient.
Honors & Awards
  • 2020: Best Innovative B-Tech Thesis Award by Indian National Academy of Engineering: Received this award for the Best Project in the Energy category. Link.
  • 2020: Winner of 3-Minute Project Presentations competition held by IEEE Kerala Chapter: Received a prize for winning the competition held by IEEE Kerala Chapter in Conscientia, technical-fest of IIST.
  • 2016-2020: Received a full scholarship for my undergraduate studies at IIST by the Department of Space, Government of India: This scholarship was awarded based on merit for bright students pursuing their bachelor's degree in space sciences and applications. It was provided by the Department of Space in collaboration with IIST.

Design and source code from Jon Barron's website.