Education

B.S. in Computer Science

University of California, Santa Barbara
06/2027

Graduate Coursework: Foundation Models (CS 291A), Robot Learning (CS 291K)

Experience

NVIDIA

ML Research Engineer Intern
06/2026 – Present | Santa Clara, CA
  • •Owned the post-training pipeline for Nemotron's long-context multimodal document reasoning, spanning synthetic data generation and SFT/GRPO, achieving strict +11% on GDP.pdf benchmark⁠
  • •Ran large-scale training experiments across hundreds of GB200/B200/H100 GPUs with Megatron-LM and NeMo RL, diagnosing model behaviors and targeting failure modes to improve omni reasoning capabilities.
  • •Open sourced all training and data recipes to NVIDIA NVFlow⁠ and released datasets to Hugging Face (coming soon).
    • 03/2026 – 06/2026 | Santa Barbara, CA
  • •Extended ChipAgents coding harness into a long-horizon, multi-agent system with persistent subagents, inter-agent communication, resumable execution, and live subagent steering (TypeScript + Bun).
  • •Built batch execution mode for reproducible, large-scale evaluation of ChipAgents on internal RTL benchmarks

    • UCSB AI Group⁠

    Undergraduate Researcher | Advisor: Xin Eric Wang
    10/2025 – Present | Santa Barbara, CA
    06/2025 – 09/2025 | San Jose, CA
  • •Wrote depthwise convolution kernels in Rust to achieve 16x faster performance on our NPU. Evaluating on MobileNetV2 demonstrated a 5.8x improvement in inference latency.
  • •Explored efficient autoscheduling for CNN/MLP operations by extending Halide to support in-memory compute primitives.
    • Awards
    10/2025

    Won the world's largest collegiate hackathon (3k+ participants, 700+ teams) building computer-use agents remotely controlled through FaceTime or iMessage. Released 1 month before OpenClaw.

    Publications

    Nemotron 3.5 Super Technical Report

    NVIDIA Research

    Dylan Lu*, Surya Gunukula*, Saaket Agashe, Tengxiao Liu, Xin Eric Wang

    Gurusha Juneja*, Dylan Lu*, et al.