XP4T.SYS  ·  REV 2026.07

Rithwik Vallabhan TV

FPGA / SoC / ASIC Engineer — Kochi, Kerala

I design and verify digital hardware — from RTL in Verilog and VHDL, through full ASIC back-end flow, to Zynq SoC deployment. Recent work spans CNN inference on FPGA fabric, AXI DMA data-acquisition pipelines, and a complete RTL-to-GDS tapeout in Synopsys ICC2 on 32 nm.

↓ SCROLL TO ENTER DESIGN TABLE

── DESIGN TABLE ──

Hover a trace to preview it — click to open the repo on GitHub.

U6 XC7Z020 ZYNQ-7000 Open OV7670-camera-asic on GitHub U1 C2 CAMERA SENSOR IF CAM CONN Open anomaly_detection_fpga on GitHub U2 C4 DPU / AI ACCEL HDMI ETH PHY Open processor_16b on GitHub U3 C6 CLK 16-BIT CPU CORE U5 DDR U4 PMIC VIN 3V3 FB1 RA1 X1 50MHz PWR STAT R3 R4 C5 R5 JTAG TCK TMS TDI TDO GND UART / PWR VIN 3V3 GND TX RX USB-C RESET BOOT TP1 TP2 XP4T BOARD · REV A DESIGNED IN KERALA XP4T LABS · 2026 FR4 · 1.6mm · 4-LAYER · HASL RoHS · 94V-0 SN:000126 · FAB 2026

── RESEARCH DESK ──

A toolbox, not a percentage bar. Pick a tool up.

Select a tool above.

── NOTEBOOK ──

Turn the page.

NIELIT Calicut

Project Engineer · Aug 2025 – Present

Delivering FPGA training and faculty development programmes under the MeitY C2S (SMART Lab) initiative. Automating course-management workflows in Python. Published work on deploying PyTorch models to Xilinx FPGAs via Vitis AI.

Indian Institute of Astrophysics

FPGA Design Intern · Feb – Aug 2025

Designed RTL IP in VHDL for CMV4000-based image acquisition on a Zynq SoC. Built a continuous high-speed capture pipeline across AXI DMA, PS DDR and BRAM, with embedded C control software in Vitis. Validated timing and signal integrity with an MSO.

Adi Shankara Institute of Engineering & Technology

B.Tech, Electronics & Communication Engineering · Nov 2021 – Apr 2025

Also served as Chairman of the Association for ECE Students (2024–25) and Quality & Operations Lead at the Innovation and Entrepreneurship Development Cell (2024).

01 / 03

── PUBLICATIONS ──

Click a paper to unfold it.

Abstract This paper presents an end-to-end workflow for deploying Artificial Intelligence (AI) and Machine Learning (ML) applications on Xilinx Zynq UltraScale+ Field Programmable Gate Arrays (FPGAs), with a focus on the ZCU104 evaluation board. Using the Vitis-AI 3.5 stack, we implement a ResNet-18 Convolutional Neural Network (CNN) trained on the Kaggle Vehicle Color Recognition (VCOR) dataset. The approach spans the entire development pipeline: configuring the host system via Docker, training the network in PyTorch, quantizing the model from floating-point to INT8, compiling the graph for the Deep Learning Processor Unit (DPU), and finally deploying the application on the target platform. We discuss practical implementation issues such as Docker setup, handling cross-compilation artifacts, and configuring IP networking. Experimental evaluation on the ZCU104 shows a maximum inference throughput of 421.05 FPS and a Top-5 accuracy of 99.0%, demonstrating the suitability of the Vitis-AI flow for edge-oriented computer vision workloads.
Keywords
FPGA
Vitis-AI 3.5
ResNet-18
INT8
DPU
ZCU104
PyTorch
Venue
CIPHER 2026
13–15 Feb 2026
NIT Jalandhar, Punjab
Added to Xplore
21 May 2026
Pages
1–6
DOI
10.1109/CIPHER70417
.2026.11523992
e-ISBN
979-8-3315-5184-1
Abstract This paper presents the design, development, and implementation of a real-time seismic-based elephant detection system aimed at mitigating human-elephant conflicts. The system leverages geophones to detect seismic infrasound signals produced by elephant movements. A low-noise amplifier conditions the weak seismic signals, which are then digitized and processed using an Arduino-based microcontroller. Advanced signal processing techniques, including spectral analysis and feature extraction, are employed to distinguish elephant rumbles from background noise such as environmental vibrations. The system provides early-warning capabilities through audible alarms and SMS alerts, offering immediate notification to authorities and local communities. The proposed detection mechanism demonstrates high sensitivity and specificity, making it an effective solution for both real-time monitoring and data collection for long-term conservation efforts. The system's low-cost, energy-efficient design makes it suitable for deployment in remote wildlife habitats, contributing to sustainable conservation practices and enhancing human safety in areas prone to wildlife intrusions.
Keywords
Seismic detection
Geophone
Infrasound
Arduino
Spectral analysis
Human-elephant conflict
Venue
ICACC 2024
6–8 Nov 2024
Kochi, India
Added to Xplore
22 Jan 2025
Pages
1–6
DOI
10.1109/ICACC63692
.2024.10845525
e-ISBN
979-8-3503-7913-6

── TERMINAL ──

xp4t.sys
$

Type help for commands  ·  ↑/↓ history  ·  tab to autocomplete

ENGINEERING SANDBOX UNLOCKED — you found the Konami code.