Aditya Ramchandra Pataki

Extended a Transformer-based model to jointly predict per-cell velocity, object detection, and occupancy for Valeo MCR2 radar, enabling end-to-end deep learning based signal processing. Developed DirectFlowMatch, a LiDAR-based scene flow algorithm for Valeo Scala2, estimating per-point velocities without object segmentation or clustering.

Investigated various compression techniques, methods grounded in statistical physics and information theory to identify the most suitable methods for LiDAR point cloud compression while retaining motion-critical features and discarding redundancies. Analyzed how objects, background, and noise affect data density and compressibility.

Integrated a pre-established cloud-based teleoperation framework into an ADAS-enabled miniature vehicle (Droid), enabling seamless real-time command execution, system monitoring, and bidirectional data exchange.

Designed industrial scrubber systems using SolidWorks and Solid Edge. Collaborated and engaged with fabricators and clients to align technical requirements.

Conducted burn-back analysis to identify propellant burning zones and optimize fuel efficiency for rocket propulsion systems. Executed experimental testing using MATLAB, Microsoft Excel, and AutoCAD to evaluate burn rates and predict rocket range and performance. Developed an interactive MATLAB-based dashboard to support rapid, data-driven decision-making.