• Developed an end-to-end Entity-Aware-LSTM (EA-LSTM) model using the HAWQS dataset, significantly outperforming hydrological benchmarks by learning catchment similarities from meteorological time series and static catchment attributes• Engineered a reinforcement learning environment utilizing the Soil and Water Assessment Tool with OpenAI's Gym framework, deploying custom deep learning policies (TD3, DDPG) to optimize crop management for higher yields and minimal input use• Built a Python-based system to interpolate 40 years of precipitation data spatially and temporally, utilizing JAX for accelerated computation and kriging with an exponential variogram to minimize MAE and analyze metrics like R-squared for improved accuracy • Developed and deployed a pipeline for real-time YOLOv8-based weed detection system using UAV images, integrating advanced data collection, preprocessing, and model optimization techniques to enhance accuracy and efficiency• Worked with IBM on developing a Large Language Model (LLM) for agriculture, focusing on precision in crop management, pest prediction, and yield forecasting using advanced NLP techniques

• Provided instructional and technical support as Teaching Assistant for the Machine Learning, Algorithms, Mobile Robotics and Noise and Stochastic Processes course, including grading, one-on-one assistance, and troubleshooting• Assisted in teaching and supporting students by providing hands-on learning with the hardware

Field Robotics & NEURAL lab under the guidance of Prof. Hanumant Singh and Prof. David Rosen• Crafted a unique implementation of bag of words algorithm and OpenGV in C++ for geometric verification in a multi-camera SLAM system, leading to enhanced mapping accuracy and robustness• Optimizing loop closure detection in a Multi Camera SLAM system through the utilization of multi-threading techniques, resulting in real-time performance in dynamic environments

• Mentored 25 first-year undergraduates, enhancing their student experience and fostering connections with support staff• Contributed to Northeastern's position as an innovator in education through leadership in mentoring and student support

• Selected to participate in Google's highly competitive Computer Science Research Mentorship Program, where I gained valuable research experience and insights into the tech industry through one-on-one and group meetings with a Google mentor and peers• Worked closely with a Google mentor and small group of peers to explore topics such as defining research problems and navigating the publication process

• Developed a Graphical User Interface (GUI) application leveraging tkinter on Python to batch process and analyze sensor data to reduce processing time and save hundreds of dollars on MATLAB licenses• Collaborated in the development of software architecture of the Insulin Delivery Device according to FDA regulations• Designed a controller leveraging ROS for Harvard Pump to improve dose accuracy testing of the insulin Delivery Device• Modelled the device on MATLAB and Simulink to examine system level performance and perform root cause• Conducted performance evaluations of the device by analyzing real-time data and directly testing it on live swine specimens

• Designed a measurement setup along with a python program for recording and displaying real-time thermal response of chargingwater in an experimental test rig using sensor column and DAQ boards• Analyzed the particle flow pattern through Particle Image Velocimetry and Background oriented schlieren on MATLAB• Innovated a universally fitting nozzle design to reduce Kelvin-Helmholtz Instabilities by up to 47% and improve stratification

• Spearheaded the team to design a highly efficient Lyophilizer with a 60% longer run-time as compared to conventional systems• Programmed PLC system to automate and shorten the development of freeze-drying cycle resulting in a lower energy consumption by the system• Conducted experiments by varying the load in the Lyophilizer; improved the cold trap (condensing coil) and the fins of the Lyophilizer by simulating results on Ansys (CFD) and MATLAB