Chiranjivan K N

🚀 Welcome! I'm a versatile computer engineer with a Master's from George Mason University.

🎓 My academic journey has equipped me with a diverse skill set, ranging from programming languages such as C, C++, C#, and Python to expertise in technologies like CUDA, VHDL, Verilog, RTL Design, Embedded Design, Azure, AWS, and Hyper-V.

💼 I finished my undergraduate studies from Sri Sairam Engg College, Chennai. After my undergrad I worked as software engineer at Vembu Technologies. At Vembu, I contributed to projects involving Hyper-V, Microsoft Azure, and Tape storage.

🛠️ My diverse experiences include FPGA-based projects, Quantum ML research, ML-based Android apps for skincare, and certifications in IBM Qiskit and NVIDIA CUDA. I have a keen interest in exploring the realms of CUDA, GPU and FPGA.

Explore my portfolio for a snapshot of my dynamic journey in computer engineering! 💻✨

Email  /  CV  /  Github

• Research
• Projects
• Course Projects

Reliable Multiplier Design Implementation on FPGA with Adaptive Hold Logic

It addressed the delay problems due to the aging effects of a transistor to increase the system performance. We used an Adaptive Hold Logic (AHL) circuit for the proper selection of cycle period and an Error Detection Correction Pulsed Latch (ECPL) for the detection of timing errors. We compared the delays of various error tolerance circuits by generating simulations among which EPCL was the lowest. In addition to it, we chose a radix-4 booth multiplier of reduced area and fanout to further improve performance. We tested our proposed architecture in Xilinx Spartan6 FPGA, and validated our implementation.

Combating Against SEM IC Reverse Engineering through Adversarial Perturbations
Chiranjivan, 2023

Developed a defense mechanism against SEM-based IC reverse engineering by introducing localized adversarial patches into logic gate GDSII. This strategy effectively fools CNNs, strengthening intellectual property protection and information security.

Adversarial Attack Evaluation using ART (adversarial-robustness-toolbox)
Chiranjivan, 2023

Evaluation of adversarial machine learning attacks on MNIST using the ART library, identifying the hyperparameters that leads to vulnerabilities, which in turn help to create strategies to enhance model robustness and security.

Dynamic Instruction Scheduling Simulator
Chiranjivan, 2023

Simulated a dynamic instruction scheduling for out-of-order processor, implementing the Tomasulo algorithm with a Reorder Buffer to optimize execution performance.

Cache Simulator
Chiranjivan, 2023

Designed and implemented a C++ cache simulator that emulates Level 1 (L1) and Level 2 (L2)cache behavior, analyzes performance metrics (miss rate, memory access time), and employs LRU eviction for various cache configurations and write policies, including tracking Valid, Dirty bits, and Tag fields.

Reliable Multiplier Design Implementation on FPGA with Adaptive Hold Logic
Chiranjivan, Pradeepkumar, PremKumar, 2019

It addressed the delay problems due to the aging effects of a transistor to increase the system performance. We used an Adaptive Hold Logic (AHL) circuit for the proper selection of cycle period and an Error Detection Correction Pulsed Latch (ECPL) for the detection of timing errors. We compared the delays of various error tolerance circuits by generating simulations among which EPCL was the lowest. In addition to it, we chose a radix-4 booth multiplier of reduced area and fanout to further improve performance. We tested our proposed architecture in Xilinx Spartan6 FPGA, and validated our implementation.

Flood Monitoring System
Chiranjivan, Pradeepkumar, 2018

Designed a system which was low cost, easy to install and highly reliable so that people living in flood prone area i.e., river/dam can be given an early warning. The flood parameters like water pressure, water level, flow, temperature are measured by the sensors and given to the ADC, which are then sent to a Raspberry Pi. Email alerts are sent the concerned authorities if the parameters of flooding are detected.

Fruit Classification Using CNN
Chiranjivan, 2022

In this project, utilized Convolutional Neural Networks (CNN) to classify fruits based on their images. The CNN model was trained and evaluated using a dataset of labeled fruit images. Additionally, I explored pruning techniques to reduce the model size for deployment on edge devices like Raspberry Pi (RPi). This project enhanced my understanding of image classification and deep learning frameworks.

Learning in SNNs, Quasi Back Propagation-Based
Chiranjivan, 2022

Focused on the challenges associated with training Spiking Neural Networks (SNNs) and explored two algorithms: Temporal Spike Sequence Learning Via Backpropagation and Spike layer Error reassignment in Time (SLAYER). These algorithms aim to overcome the limitations of traditional backpropagation methods when dealing with the non-differentiability of discrete spike events in SNNs. By comparing the accuracy results on standard datasets, analyzed the performance of these algorithms and their effectiveness in learning temporal spike sequences. This project provided valuable insights into the complexities of SNN training and contributed to ongoing research in neural networks.

Designing Hyperdimensional Computing Systems with FPGA Technology
Chiranjivan, 2023

Implemented Hyperdimensional Computing (HDC) on Field-Programmable Gate Arrays (FPGAs). HDC is a promising computing paradigm inspired by the brain's information processing abilities. By representing data as high-dimensional vectors, HDC offers state-of-the-art solutions in tasks like image classification and natural language processing. Designed an HDC inferencing framework specifically for the FPGA platform, analyzed resource utilization and accuracy metrics. This project demonstrates the feasibility and potential of implementing HDC on FPGA platforms.

Big Data Analysis of Anime
Chiranjivan, 2023

Conducted a study on big data analysis of anime using data from topAnime.org. Utilized Spark libraries for Python to process and analyze a large dataset comprising over a million user reviews and ratings of anime series. The study uncovered valuable insights into anime viewer preferences, popular genres, highly rated series, and the correlation between user reviews and ratings. These findings highlight the utility of Spark libraries in big data analysis for understanding audience preferences and tailoring content. The techniques employed in this project have broader applications in analyzing user behavior and preferences across various media forms.