LUKE SHIMANUKI luke@shimanuki.cc | https://luke.shimanuki.cc ==================================================================================================== EXPERIENCE ---------------------------------------------------------------------------------------------------- Aurora Innovation | Senior SWE / Tech Lead (Behavior ML) 2022–present - Led team of 7 ML engineers to develop & deploy forecasting models, reducing error by 20% on targeted scenarios, resolving 100+ behavioral test failures across 7 safety-critical failure modes, & enabling driverless operations that achieved 100K+ incident-free miles. - Recognized by senior leadership for urgently resolving intersection behavior deficiencies blocking 2 investor showcases central to $800M+ funding round & manufacturing partnership agreement. - Mentored 3 ML engineers, & developed framework accelerating integration of separate model improvements from 1 week to 2 days, enabling the team to double in size while maintaining development velocity. Magna Electronics (formerly Optimus Ride) | SDE II → Senior SDE (Motion Planning) 2020–2022 - Led 3 engineers to develop probabilistic uncertainty quantification framework & risk-aware motion planner, enabling safe autonomous navigation through contested intersections. - Led cross-functional initiative to redesign prediction-planning interface incorporating probabilistic uncertainty propagation, facilitating risk-aware planning across the autonomous driving stack. - Formulated technical requirements & algorithmic specifications for 2 ADAS features following ISO 26262 functional safety standards, resulting in 1 patent filing. Cruise Automation | Intern (Prediction) Summer 2018 - Improved forecast dynamic feasibility & intersection reasoning, tripling passing test cases & substantially reducing intersection disengagements. Optimus Ride (Self-driving) | SWE Intern (Perception) Summer 2017 - Designed & implemented end-to-end perception pipeline from object detection through tracking & forecasting, establishing architecture that integrated with motion planning system. RightHand Robotics (Robotic Grasping) | Intern (Machine Learning / Vision) January 2017 - Designed & trained CNN-based grasp prediction model from RGB-D input, improving pick success rate on cluttered bin-packing scenarios. Tanius Technology (Prop Trading) | Developer (ML Modeling) 2015–2016 - Built end-to-end ML system for algorithmic trading: data ingestion, model training, & backtesting infrastructure. EDUCATION & RESEARCH PUBLICATIONS ---------------------------------------------------------------------------------------------------- Massachusetts Institute of Technology | CSAIL Learning & Intelligent Systems 2016–2020 M.Eng Electrical Engineering & Computer Science (AI Concentration), GPA 5.0 (out of 5.0) S.B. Computer Science + Brain & Cognitive Science, GPA 4.9 (out of 5.0), 5.0 within CS major B. Axelrod*, L. Shimanuki*, "Efficient Motion Planning under Obstacle Uncertainty with Local Dependencies", WAFR 2022 L. Shimanuki*, B. Axelrod*, "Hardness of Motion Planning with Obstacle Uncertainty in 2 Dimensions", IJRR 2021 B. Kim*, L. Shimanuki*, et al., "Representation, learning, and planning algorithms for geometric task and motion planning", IJRR 2021 B. Kim*, L. Shimanuki*, "Learning value functions with relational state representations for guiding task-and-motion planning", CoRL 2019 L. Shimanuki*, B. Axelrod*, "Hardness of Motion Planning with Obstacle Uncertainty in 3 Dimensions", WAFR 2018 * indicates equal contribution SKILLS ---------------------------------------------------------------------------------------------------- Languages: C, C++, Python ML/AI: PyTorch, TensorFlow, ROS Tools: Git, Docker, Linux ==================================================================================================== ACTIVITIES ---------------------------------------------------------------------------------------------------- Site Manager | Food for Free COVID-19 Relief Program | 2020 Directed team of 12 volunteers for packing & delivering groceries to ~300 households weekly. Program Director, Head Webmaster | MIT Educational Studies Program | 2017–2020 Directed educational programs reaching ~3000 MS/HS students, with ~1000 classes taught by ~500 teachers, & run by ~100 volunteers. Mentored 5 future program directors. Maintained website used by ~5000 students. Software Lead | AVBotz | 2012–2016 Led the software team (~12 members) for fully-autonomous submarine for manipulating objects, aiming & shooting torpedoes, & navigating around obstacles. International finalist (7th Place) at RoboSub 2015. Co-President, HPMS Branch Director | ACE Coding | 2013–2016 Managed ~16 volunteers to teach weekly programming lessons to ~100 middle school students annually. Organized full-day event attracting ~300 students. Taught machine vision & ML workshops. AWARDS ---------------------------------------------------------------------------------------------------- USACO Platinum | Intel STS 2016 Semifinalist | Eagle Scout | MIT Battlecode 2018 Finalist PROJECTS ---------------------------------------------------------------------------------------------------- C++: Low latency audio streaming to enable remote musicians to play in-sync with under 30ms latency Python: Musical autocomplete to assist chord & melody composition Python: Open-source window manager extension enabling grid layout C: Web browser using Chromium's rendering engine with configurable vi-like key bindings C: C compiler to convert C code to x86 assembly Java: Neural network AI for a multiplayer platformer fighting game n/a: Inflatable Sailboat: DIY spritsail rigged to an inflatable raft