Skip to content

SIDDHARATH SHAKYA

@SidXpro

Skill iconPython
Skill iconJavaScript
Node.js
Skill iconReact.js

Mandi, India

πŸ‘‹ Hey Devfolio! I'm Siddharath Narayan Shakya

I'm a deep learning researcher, hardware hacker, and AI enthusiast currently working as a Research Assistant at IISc Bangalore. I love building real-world AI systems β€” from FPGA-based crypto hardware to intelligent compression pipelines and robotic controllers using VLMs.


πŸ’‘ About Me

  • πŸŽ“ M.Tech (Signal Processing), IIT Mandi
  • 🧠 Research in Deep Learning, Hardware Security & Robotics
  • βš™οΈ Tools I use: PyTorch, Verilog, Blender, ROS, HuggingFace, PYNQ Z2, and more
  • πŸ“ Based in India | Open to collaboration | Always building

πŸ› οΈ Core Skills

  • Vision + Signal Processing with Transformers & CNNs
  • Secure & Lightweight Cryptography (ASCON, KECCAK, PUFs)
  • FPGA/ASIC Design for AI Acceleration
  • VLM-powered Robot Planning (ReKep, RoboMimic, LIBERO)

πŸš€ Hackathon Highlights

  • 🎡 2nd Runner-Up – FrostHack 2024, DevFolio
    Generated AI-based music using real-time mood detection with facial cues

  • πŸ₯‰ 4th Global Rank – DCASE 2024 Challenge
    Environmental sound scene synthesis using latent diffusion models
    Technical Report

  • πŸ₯‡ 1st Rank – National Space Day Hackathon 2024
    Built a 3D immersive VR simulation of the solar system using WebGL + JS


πŸ“š Publications

  • Deep Learning Based Compressive Domain Analytics for Seismic Images
    Accepted at ICVGIP 2024, IIIT Bangalore | ACM Proceedings
    DOI: 10.1145/3702250.3702290
    πŸ”Ή Proposed a deep compression framework enabling accurate fault line detection at low bitrates.

  • Deep Priors for Video Quality Prediction
    Accepted at ICVGIP 2024, IIIT Bangalore
    Archive
    πŸ”Ή Leveraged learned priors for efficient perceptual quality prediction in videos.

  • A Novel Algorithm for Elective Allocation in CBCS
    Published in Springer AISC, TCEC Conference 2020
    DOI: 10.1007/978-981-15-0058-9_5
    πŸ”Ή Designed an optimized algorithm to allocate subjects fairly using ML and data structures.


πŸ” Recent Projects

  • πŸ” ASCON + KECCAK-p on FPGA
    Implemented a quantum-secure authenticated cipher for IoT security

  • πŸ›°οΈ Model Predictive Control for Satellite Detumbling
    Built a GMRES-based MPC pipeline using magnetic torque-only actuation

  • πŸ€– Robotic Manipulation using ReKep + VLMs
    Integrated Robomimic + ReKep for task-conditioned robotic planning

  • 🧠 Entropy-Aware Compression for Seismic & Medical Imaging
    Achieved better SSIM/PSNR using learned priors and attention-based quantizers


⚑ What Drives Me

I thrive at the intersection of AI, hardware, and systems design. I believe in building intelligent, secure, and efficient solutions that translate well into the real world. My goal is to combine research with rapid prototyping to solve meaningful problems β€” whether it's in space, healthcare, robotics, or security.


πŸ“« Let’s Connect


πŸ›  Thanks for reading! I'm excited to build, collaborate, and innovate at this hackathon. Let’s create something impactful! πŸš€