• Designed and iterated on multi-modal perception models combining 2D object detection and semantic segmentation, progressing to end-to-end transformer-based architectures with direct 3D bounding box prediction.

• Optimized model latency for on-vehicle deployment through architecture redesign, bottleneck profiling (GPU flame graphs), and quantization via TensorRT - balancing detection accuracy against real-time safety-critical inference requirements.

• Helped architect and manage the internal training framework (PyTorch Lightning wrapper) - became a go-to engineer for model training issues including data throughput, architecture sizing, and optimization bottlenecks.

• Evaluated and integrated state-of-the-art research into production perception models - reading, benchmarking, and adapting novel architectures to meet on-vehicle compute and latency constraints.

• Verified and validated perception models for safe on-vehicle deployment, ensuring compliance with autonomous vehicle safety cases.

• Named inventor on 3 granted U.S. patents spanning scene embedding-based detection, construction zone perception, and collision avoidance trajectory prediction - covering sensing through planning across the full autonomy stack.

• Implemented structure-from-motion and 3D reconstruction algorithms in C++ - converting raw mobile video into navigable 3D meshes using off-the-shelf libraries (COLMAP/Ceres) with significant custom optimization.

• Identified and eliminated critical performance bottlenecks in the reconstruction pipeline.

• Designed and productionized backend reconstruction pipelines on AWS - architecting job orchestration, storage optimization, and data handoff to the frontend team for a consumer-facing product.

• Operated across the full backend stack in a startup environment - spanning CV algorithm development, cloud infrastructure design, and pipeline reliability beyond core computer vision responsibilities.

• Designed object detection models optimized for extreme latency constraints - running inference directly on security camera hardware with no dedicated GPU, pioneering early edge AI deployment patterns.

• Helped develop the company's in-house deep learning framework in C++

• Worked on intelligent caching strategies for neural network inference - leveraging temporal scene consistency to skip redundant computation and dramatically reduce per-frame processing cost on edge devices.

Worked with Professor Sergey Levine on deep reinforcement learning for autonomous robotic skill learning, focusing on model-based updates for improved sample complexity.

Developed in-house APIs for production with Node.js/Postgres/Redis. Helped design/build new business products with Node.js. Helped debug issues arising in production.