🎓 MS AI @ USC | 🤖 Robotics Research @ USC ISI | ⚡ Incoming Failure Analysis Data Intern @ Bloom Energy

Physics-informed ML engineer applying deep learning and optimization to real-world physical systems, from modular robots to fuel cells to nuclear reactors. I build intelligent systems that bridge the gap between simulation and hardware in energy, robotics, and autonomous systems.

@ Bloom Energy (Summer 2026)

• Applying ML to solid oxide fuel cell sensor telemetry for anomaly detection and failure analysis
• Time-series analysis and root cause diagnosis on high-temperature electrochemical systems

@ USC Information Sciences Institute — Polymorphic Robotics Lab

• Lead software engineer on distributed inverse kinematics for autonomous multi-robot coordination
• Simulation-to-hardware validation pipeline deploying algorithms on physical robots
• Engineered closed-loop Inverse Kinematics (IK) solvers utilizing real-time physics readback to counteract gravity-induced joint sag in multi-link modular systems
• Optimized high-fidelity physics constraints in Unreal Engine 5 by correcting centimeter-based inertia calculations, achieving critical damping for reconfigurable robot modules

@ Lineslip Solutions

• Production RAG pipelines serving 10K+ queries/day with sub-second latency
• 40% performance improvement through algorithmic optimization and quantization

Applying AI and optimization to hard tech problems where physics constrains the solution:

• ⚛️ Nuclear Energy & Clean Power — Reactor modeling, digital twins, predictive maintenance, fault diagnosis
• 🔋 Energy Systems — Fuel cell degradation analysis, battery optimization, grid reliability
• 🤖 Robotics & Autonomy — Distributed control, perception, multi-agent coordination
• ✈️ Aerospace — GNC, propulsion optimization, eVTOL systems

Deep learning on time-series reactor sensor data for accident detection and diagnosis

• Semi-supervised anomaly detection on 96 operational parameters across 18 accident scenarios
• Autoencoder, LSTM, and Transformer architectures achieving 0.985+ AUROC
• Multi-class accident classification with SHAP-based root cause attribution
• Physics-informed digital twin surrogate model
• Data: NPPAD dataset (Nature Scientific Data) from PCTRAN PWR simulator
• Tech: PyTorch, pandas, scikit-learn

Production-scale video processing achieving 6.45 FPS on 1080p video

• RAFT optical flow for temporal consistency and ego-motion estimation
• Trained on 118K MS-COCO images using distributed PyTorch DDP with custom CUDA kernels
• 30% training speedup through CUDA optimization
• Tech: PyTorch, CUDA, RAFT, Computer Vision

Distributed inverse kinematics for autonomous multi-robot coordination (USC ISI)

• 96% tracking accuracy with sub-50ms real-time latency
• Multi-modal sensor fusion pipeline (camera + IR)
• Simulation-to-hardware deployment on physical robots
• Tech: Python, C++, ROS, Unreal Engine, MuJoCo, NumPy, SciPy

AI-powered coaching with fine-tuned Llama 3.2 LLM — Won 55th Annual Senior Design Conference

• QLoRA PEFT for culturally-aware content generation
• Full-stack web application with goal tracking
• Tech: Python, Typescript, Llama, QLoRA, Firebase

Core: Distributed Optimization · Sensor Fusion · Time-Series Analysis · Anomaly Detection · Physics-Informed ML

Languages: Python · C/C++ · CUDA · MATLAB · JavaScript

ML/AI: PyTorch · TensorFlow · Computer Vision · LLMs · scikit-learn

Robotics: ROS · Sensor Fusion · Path Planning · Multi-Agent Systems

Production: FastAPI · Docker · CI/CD · Elasticsearch · Git · Linux

• 📝 First-author publication at AHFE Hawaii 2024 — AI-facilitated creative interfaces
• ⚛️ Built ML fault diagnosis system for nuclear reactor sensor data (NPPAD/PCTRAN)
• 🤖 96% tracking accuracy on distributed multi-robot coordination (USC ISI)
• 🎬 Trained neural style transfer on 118K images with distributed GPU training
• ⚡ Production ML systems serving 10K+ queries/day with 40% performance gains
• 🥇 55th Annual Senior Design Conference Session Winner — Santa Clara University

Creative Collaborator: AI-facilitated UI for Creating Engaging and Insightful Memes First Author | AHFE Hawaii 2024 | DOI: 10.54941/ahfe1005579

• 💼 LinkedIn
• 📧 rjnickel@usc.edu
• 📍 Los Angeles, CA