Sacharitha Sirisilla

Background

About Sacharitha Sirisilla

Hei! I am a Computer Science Master's student specializing in Data Science and AI at Kancheepuram, currently in my final semester. As part of an exchange program, I am completing my thesis at the University of Agder (UiA), Norway, where I also work as a research assistant. My expertise lies in machine learning and deep learning, computer vision, with hands-on experience in Transformers, object detection and multi-class classification algorithms, and, more recently, Tsetlin Machine. I also have a strong foundation in GenAI, Large Language Model (LLM) applications, and Natural Language Processing (NLP). I am proficient in Python, C, and C++, with a solid understanding of data structures and algorithms (DSA). Additionally, I have a strong grasp of mathematics and statistics, essential for data science and analysis, and I have extensive experience in data preprocessing and exploratory data analysis (EDA). At UiA, I have worked with diverse datasets, including CSI and RSSI signals from drones, human activity recognition data collected through accelerometers and gyroscopes, and LiDAR data for vineyard crop health monitoring. As part of my master’s thesis at UiA, I developed a robust and distributed real-time human intruder detection and tracking framework using multiple AWR1843 FMCW radars. The system introduced a self-tunable space-time DBSCAN algorithm that adapts clustering parameters based on average pairwise distance (APD), mitigating issues of density variation and clutter. I extended this with dual-mode clustering (HDBSCAN + OPTICS) and ensemble spectral clustering for spatial segmentation across diverse viewpoints. For temporal association and ID consistency, I used the Hungarian algorithm, combined with Kalman filtering for trajectory smoothing. Each radar node, paired with a Raspberry Pi 4, performed edge-level processing and communicated with a central ROS-based fusion server. Using Sampling Covariance Intersection (SCI), local tracks were fused into globally consistent trajectories. This low-latency, scalable system demonstrated robust tracking performance under occlusions and dynamic multi-person scenarios, pushing the boundaries of autonomous surveillance through sensor-level intelligence and real-time decision-making. I am actively seeking full-time opportunities in Data Science, Machine Learning, or AI Engineering roles. I am also open to research positions and PhD opportunities in the fields of ML/AI, computer vision, and sensor fusion. Please feel free to connect if you'd like to collaborate or discuss opportunities.