李君翔

Key Skills

Software

Other

Top Subject Matter

Autonomous driving / vehicle dynamics (off-road reinforcement learning control)

Engineering document translation / localization (layout-preserving)

Automotive engineering literature analysis and technical content preparation

Top Data Types

Document

Text

Top Task Types

RLHF

Freelancer Overview

Off-Road Adaptive Control with Reinforcement Learning and CarSim Co-Simulation. Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal, Proprietary Tooling, and PDFMathTranslate-next. Education includes Bachelor of Science, N/A (2022). AI-training focus includes data types such as Computer Code, Programming, and Document and labeling workflows including RLHF, Translation, and Localization.

Labeling Experience

Vehicle Body Structure Literature Analysis and Technical Content Preparation

OtherDocument

Conducted technical document processing including English-to-Chinese translation support and organization of automotive engineering literature. Prepared structured technical content for downstream academic or presentation use by synthesizing key insights from multiple papers. Supported reading, translation interpretation, and content organization around vehicle body structure and crashworthiness topics. • Reviewed and organized technical literature on vehicle body structures and crash performance. • Provided translation and interpretation support for automotive engineering papers from English to Chinese. • Prepared structured technical content suitable for academic writing and presentations. • Helped extract engineering insights on structure design, crash energy absorption, and body optimization themes.

2025 - Present

Engineering PDF Translation Workflow with Layout Preservation

Document

Developed a layout-preserving workflow to translate technical engineering PDFs from English to Chinese for high-fidelity document output. Used LLM-backed translation services with configurable backends and runtime settings to support reliable translation of long technical documents. Performed troubleshooting to address local configuration and model/runtime stability issues during execution. • Set up a reproducible translation workflow for born-digital engineering PDFs with layout preservation. • Configured and switched between DeepSeek and OpenAI-compatible backends based on task needs. • Produced bilingual and monolingual translated PDFs while maintaining high layout fidelity. • Built an operational process for debugging translation failures and handling long-document generation.

2025 - Present

Off-Road Adaptive Control with Reinforcement Learning and CarSim Co-Simulation

RLHF

Built and validated an RL training workflow for off-road adaptive control using a DDPG-based approach within a Python–CarSim co-simulation environment. Trained and monitored model behavior by designing reward and metric pipelines tied to vehicle dynamics states and control outputs. Performed iterative debugging and simulation alignment to ensure training traces and downstream visualization were consistent with vehicle response objectives. • Trained and refined DDPG logic for speed tracking and traction-oriented control behavior. • Designed/collected training metrics including reward, actor/critic losses, and step-wise analyses (speed, slip, traction efficiency, yaw rate). • Supported model export to enable downstream co-simulation validation and result visualization. • Coordinated simulation I/O mapping between Python and CarSim to align signals and generate test traces.

2025 - Present

Education

Work History