Resume

Ph.D., Mechanical Engineering / Computer Science

• Dissertation research on multi-modal generative models and mechanism design; MoE transformers, VAEs, large-scale training on HPC.
• GPA: 3.71

M.S. in Mechanical Engineering

• GPA: 3.86

B.S. in Mechanical Engineering

• GPA: 3.81

Robotics Engineer

• High-fidelity Isaac Sim / Isaac Lab (PhysX) workflows for URDF/CAD robots — accurate kinematics, dynamics, and collisions across dual-arm + gripper and mobile platforms.
• Tuned contact, friction, restitution, and solver settings for stable contact-rich and deformable manipulation.
• Vectorized simulation pipelines for grasps, expert demos, and large-scale imitation / policy-training datasets.
• Integrated cuRobo for collision-aware planning in PhysX environments.
• Fine-tuned π0.5 with domain-randomized sim data; suction end-effectors and deformable objects (cloth, bags).
• Blender CAD/STL prep (STL→OBJ, materials, cleanup) for sim-ready assets.
• VR teleoperation to real robots for data collection; ROS integration where applicable; VR/desktop control for arms and hand / parallel grippers.

Robotics Engineer Intern

• Built large-scale synthetic datasets in Isaac Sim for SO-100 and Trossen arms (IK rollouts, domain randomization, multi-sensor); datasets on Hugging Face (1k+ monthly downloads).
• End-to-end Omniverse workflows — scene/physics, USD stages, CAD/URDF, cameras, Replicator auto-labeling.
• Simulation tasks for manipulation, grasping, and pick-and-place before hardware deployment.
• Fine-tuned Gr00t and π0 in PyTorch on synthetic and real data; closed-loop testing on physical SO-100 with 90–100% task success on benchmark routines.

Graduate Research Assistant

• Designed and trained multi-modal LLM-style models with Mixture-of-Experts (MoE) for path synthesis, built from scratch with a hybrid ViT + decoder architecture; CLIP-style contrastive learning and Gaussian soft objectives, with ~15% accuracy gains over strong baselines at ~200 ms inference latency.
• Improved training efficiency with Classifier-Free Guidance and LoRA (~30% compute reduction); gradient tracking, adaptive clipping, and CLIP-aligned objectives for cross-modal consistency.
• Developed β-VAE and graph-based VAE models for structured latent representations supporting multi-modal generation and downstream reasoning.
• Co-authored peer-reviewed work on a 3M-sample planar linkage dataset for ML-driven path synthesis (ASME JMD) and on conditional β-VAE path generative models for mechanism design (ASME JMR).
• Compared multiple curve representations (Fourier descriptors, wavelets, point coordinates, images) inside unified generative frameworks for four-bar coupler synthesis.
• Led end-to-end ML work: curation and preprocessing of 12M+ multi-modal samples; distributed training of sub-1B transformer models on SLURM-managed HPC; scaling experiments on cloud GPUs; 200+ experiments tracked with PyTorch Lightning, Hugging Face, and Weights & Biases.

Computer Vision Engineer

• Fine-tuned YOLOv11 toward ~100% detection accuracy; cut latency by replacing a two-step pipeline with a single optimized model.
• Automated myCobot 280 PI capture (~60% less manual work, ~3× labeling throughput) with AWS S3 uploads.
• Real-time iPhone QR detection (SwiftUI + CoreML), Docker-ized for deployment.
• Real-time screen/display classification on iOS: CLIP ViT-L/14@336 + MLP (YOLO insufficient for reliable screen boundaries in the wild); WebSocket streaming, 5-crop TTA, EMA, 3-frame hysteresis for stable live inference.
• Training utilities — albumentations, batched GPU feature extraction, stratified splits, disk caching (scikit-learn, PyTorch).

Mechanical Engineer Intern

• Modeled wind turbines' interactions with rooftops through advanced simulations, enhancing reliability and performance.
• Conducted comprehensive cost-benefit analysis for data-driven project profitability and sustainability decisions.
• Validated structural integrity of designs with rigorous wind load simulations using Ansys.
• Designed novel rooftop installations for wind turbines with Autodesk Inventor CAD.

Summer Mechanical Engineer Intern

• Supported onshore operations for seamless project execution with contractors.
• Materials management, coordination, and standards compliance.
• Oversaw valve sizing, certification, and repairs to enhance efficiency and safety.
• Led piping design and stress analysis to optimize installations and improve structural integrity.

Undergraduate Research Assistant

• Explored 3D printed part anisotropy with size effect methods, enhancing materials science research.
• Created accurate test specimens via advanced 3D printing and modeling.
• Performed detailed 3-point bending tests using Instron equipment and analyzed data.

• Dean's List

• Pasha Hackathon 4.0 Winner
• TransitHack 2.0 Hackathon Winner

• Junior Research Award — International Association for Shell and Spatial Structures (IASC)