Swagat Das

Developing scalable deep-learning pipelines for semantic segmentation of large aerial LiDAR point clouds in geospatial and remote-sensing applications.

• Built end-to-end workflows for preprocessing, training, inference, and evaluation across large 3D point-cloud datasets.
• Used GPU and HPC environments for distributed batching, mixed-precision training, and high-volume data processing.
• Addressed class imbalance, sensor variation, and cross-region generalization through sampling, loss design, domain adaptation, and hard-example analysis.
• Developed a scalable stack spanning ETL, model experimentation, inference, and quantitative error analysis.

Developing a real-time traffic perception pipeline for edge deployment using NVIDIA Jetson devices and real CCTV video.

• Integrated YOLO-based object detection and drivable-area segmentation with ByteTrack multi-object tracking.
• Developed road-bound filtering, stable vehicle identities, class voting, trajectory histories, and dwell-time analytics.
• Evaluated performance under occlusion, low-confidence detections, fragmented tracks, and duplicate-track conditions.
• Designed the workflow for resource-aware edge inference and intelligent transportation applications.

Building a phased smart-manufacturing cybersecurity testbed that converts structured plant descriptions into reproducible cyber-range environments.

• Generates plant digital twins from SysML-based system representations and network graphs.
• Represents assets, connectivity, process relationships, and operational context for structured system reasoning.
• Converts generated environments into containerized Docker and Containernet cyber ranges.
• Supports attack-graph generation, controlled vulnerability testing, and red-team/blue-team evaluation.
• Integrates human-supervised agentic LLM workflows for bounded tool use, evidence tracking, review, and traceable recommendations.

Contributing to diffusion-based research in 2D and 3D medical imaging and anatomy-aware prediction for radiotherapy applications.

• Working with conditional diffusion models and U-Net-based denoising architectures.
• Integrating volumetric CT representations with slice-level prediction to preserve global 3D context.
• Using pretrained segmentation features and cross-attention for anatomy-aware conditioning.
• Evaluating image similarity and clinically relevant quantitative measures across multiple treatment settings.

A multi-institutional R&D effort focused on reliability, observability, and operational resilience of EV charging infrastructure.

• Built scalable pipelines aggregating approximately one million charging sessions into state, city, and charger-level features.
• Designed a charger health score from availability, failures, retries, utilization, and temporal weighting.
• Supported short-term forecasting using spatial and temporal relationships across charging locations.
• Co-developed a Dash and Mapbox platform with KPI cards, drilldowns, trend views, forecasts, and diagnostics.
• Improved responsiveness using Dask-based parallelization, caching, and decoupled data and interface components.

Designed an early-stage AI pipeline for ECG image classification and anomaly detection.

• Combined neural style transfer with self-supervised representation learning.
• Explored deep architecture design, pretext tasks, and optimization using PyTorch and TensorFlow.

• Designed and analyzed space-vector modulation schemes using MATLAB and Simulink.
• Compared total harmonic distortion, current ripple, and switching-state utilization.

• Reviewed substation components, protection requirements, and bus configurations.
• Studied flexibility, communication, and reliability requirements for future smart-grid substations.

• Built basic medium- and low-voltage network models in OpenDSS.
• Interfaced OpenDSS with Excel VBA and MATLAB for analysis and automation.

• Assessed harmonic distortion in an electric power system.
• Designed a passive filter in PSCAD to reduce voltage distortion toward IEEE 519 limits.

• Designed a 480 kW grid-interactive solar installation with expansion capacity.
• Used HelioScope for system design and SAM-PV for economic analysis.

Machine Learning, Deep Learning, Self-Supervised Learning, Diffusion Models, Generative AI, Transformers, Graph Neural Networks, Anomaly Detection, Time-Series Forecasting, Predictive Modeling, Domain Adaptation, Class-Imbalance Handling, Explainable AI, Responsible AI, Human-AI Collaboration

LiDAR, Point-Cloud Processing, 2D and 3D Perception, Object Detection, Semantic Segmentation, Instance Segmentation, Multi-Object Tracking, Video Analytics, Trajectory Analysis, CT Image Analysis, Medical Imaging, OpenCV, YOLO, ByteTrack

NVIDIA Jetson, Embedded Vision, Embedded Systems, Python, C++, PyTorch, Pandas, NumPy, Dask, cuDF, GPU Computing, High-Performance Computing, Parallel and Distributed Computing, Linux, Git, Jupyter

Cyber-Physical Systems, Industrial Cybersecurity, Operational Technology, Industrial Control Systems, SCADA, Smart Manufacturing, Digital Twins, SysML, Attack Graphs, CAN Bus Security, Intrusion Detection, Adversarial Machine Learning, Vulnerability Assessment, Cyber-Range Evaluation, Red-Team/Blue-Team Evaluation, NIST Cybersecurity Framework

Large Language Models, Agentic AI, Human-Supervised AI, Tool-Augmented Reasoning, Model Context Protocol, Knowledge Representation, Graph-Based Reasoning, Human Oversight, Docker, Containernet, Dash, Plotly, Mapbox, Geospatial Analytics, MATLAB

• Developed a real-time EV charger health analytics and forecasting platform with Idaho National Laboratory using approximately one million charging sessions.
• Developed scalable aerial LiDAR and 3D point-cloud segmentation pipelines using PyTorch, GPU/HPC, semantic segmentation, domain adaptation, and class-imbalance handling.
• Developed an edge-based traffic video analytics pipeline on NVIDIA Jetson devices using YOLO, ByteTrack, road-bound filtering, and trajectory analysis.
• Building a phased smart-manufacturing testbed spanning SysML and network-based digital twins, Docker/Containernet cyber ranges, attack graphs, and controlled vulnerability evaluation.
• Developing self-supervised anomaly detection and diffusion-based 2D/3D medical-imaging methods with emphasis on explainability, responsibility, and human oversight.

• Led accreditation, ranking, audit, compliance, and institutional reporting involving NAAC, NBA, ABET, QS, THE, and NIRF.
• Built data-driven academic assessment and sustainability dashboards.
• Coordinated evidence, reporting, and communication with national and international external bodies.

• Taught and developed 12+ undergraduate and graduate courses in power systems, power electronics, renewable energy, IoT, and industrial automation.
• Helped establish industrial automation and PCB manufacturing laboratories with industry partners.
• Contributed to curriculum development, applied research, accreditation, project supervision, and student mentoring.
• Supported international students and institutional internationalization activities across a diverse campus community.

• Conducted research on power-semiconductor aging, condition monitoring, and reflectometry-based fault diagnosis.
• Designed and built an accelerated aging and diagnostic test platform using regulated power delivery, thermal management, PCB gate drivers, simulation, and laboratory validation.
• Worked with MATLAB, Simulink, PSpice, LabVIEW, ANSYS, circuit instrumentation, and high-power switching hardware.

• Introduced students to testing and measurement instruments, laboratory safety, experimental methods, and data interpretation.
• Supported hands-on laboratory work connecting circuit theory with physical measurements.

• Received practical exposure to substations, transformers, CTs, PTs, circuit breakers, protection systems, transmission lines, and maintenance practices at OPTCL.
• Studied transformer design, manufacturing, testing, and industrial documentation at ALFA Transformer.

• IEEE Industrial Electronics Society Students and Young Professionals Activity Award, 2025
• Dean's International Scholar Award, University of Missouri–Kansas City, 2014
• Rajya Puraskar

• Institute of Electrical and Electronics Engineers (IEEE)
• IEEE Industrial Electronics Society
• IEEE Power and Energy Society
• Institution of Engineering and Technology