Tilak Marupilla
I am a graduate student at Northeastern University, Boston. I am currently working as a Research Assistant at WiNES Lab exploring various Generative AI and Reinforcement Learning algorithms and their implementation on FPGA and microcontrollers. My experience includes Embedded Software, Hardware Design, DSP, Wireless Communications, Deep Reinforcement Learning, and Cryptography.
Disclaimer: Most of the images and reference research paper PDF links included below are not owned or authored by me. Full credit is given to their respective creators.
Implementation of a Generative Transformer Architecture on a Xilinx FPGA using Vitis AI
This project involves implementing a transformer architecture on FPGA hardware using Vitis AI, with a focus on performance acceleration and efficiency improvements.
Deep Reinforcement Learning-based resource scheduling for 5G MAC scheduler (Rice University and ITU)
Participating in the ITU-Rice Summer 2024 competition, optimizing an existing reinforcement learning algorithm for improved 5G network efficiency.
NL2SQL â A language to SQL translation on a small device language model
Developing a small language model optimized for ARM architecture to efficiently translate natural language queries into SQL on small devices.
Presented a protocol for enhancing the security of digital amateur radio satellite systems, in collaboration with the Open Research Institute.
Proposed a technique for detecting water vapor indoors using 60GHz Wi-Fi technology, presented during the Sony India Research contest. Later in my graduate study I found that the proposed technique works at 100GHz and not at 60GHz.
âPrimer to Wyners wiretap codesâ, South Indian SDR group conference invited speaker, May 2023.
Introduced Wyners wiretap codes and their application in secure communications at this invited session.
Research Assistant - Northeastern University, Boston (July 2024 - Present)
Working on RF and AI training using ray tracing, GPU-based data processing, and developing segmentation models for peak loss prediction. Exploring the use of Gen AI models for wireless channel data prediction.
Embedded Software Engineer - Saankhya Labs Pvt. Ltd (Dec 2021 - May 2023)
Co-developed firmware modules for ARM-based SDR radio chips and conducted RF testing and simulations for improved performance.
R&D Consultant - Fortytwo Labs Pvt. Ltd (May 2018 - March 2021)
Architected secure IoT systems, proposed a QKD system, and developed secure mobile solutions leveraging ARM TrustZone. Built many prototypes for Drones and secure storage.