As a PhD student at University of Michigan, I'm primarily working in the computer architecture area under the guidance of Prof. Reetuparna Das.
My Undergradute Research was in the following ares: Computer Architecture, Memory Systems, Formal Verification(Runtime Verification), Embedded Systems, Digital Systems, Compilers, Hardware Security, Machine Learning, Deep Learning, VLSI, Robotics, CNN
Exploiting Criticality using Hardware Prefetching and Branch Prediction Techniques
Not all load accesses matter for the performance of core. Only those accesses that lie on the critical path of execution affect the core performance. Exploiting Criticality using Hardware Prefetching and Branch Prediction Techniques to enable better performance of systems. Integrating other Memory Technologies to L1 Level Cache.
Runtime Verification
Formal method to analyse computing systems based on the information from a running system and to detect the satisfiability of High level Specifications. It has many applications in aerospace and automation.
Multi AI Core Architectures
Optimizing multi-AI core architectures involves the utilization of prefetchers and dead block predictors, with Verilog serving as the implementation language. Profound insights have been acquired into the functionality and application of advanced tools such as SMAUG and gem5-aladdin during this optimization process.
Understanding the Advancements in Cache Replacement Policy and Cache Partitioning Techniques
A survey of different cache policy on different workloads to see the profile workloads into different categories. Implemented UCP and Hawkeye Predictor on ChampSim simulator.
Stock Price Prediction using ML
Exploring machine learning techniques that extend beyond relying solely on historical prices. Further expansion is being carried out to propose strategies that will indicate how effective small changes can be in financial markets. Analysis of volatility of different instruments will be carried out subsequently.
ML/DL Oriented Vegetative Drought Prediction
Our research project focused on leveraging machine learning to improve drought prediction accuracy. We explored various machine learning algorithms, create prototypes, and evaluate their real-world practicality to revolutionize water resource management and enhance drought prediction.