Bryan M.

Key Skills

Software

AWS SageMaker

Don't disclose

Other

Top Subject Matter

Autonomous driving lane detection for camera-based perception

ML platforms: distributed training jobs and deployment drift detection

Reinforcement learning for autonomous drone navigation

Top Data Types

Image

3D Sensor

Text

Top Task Types

Segmentation

Fine-tuning

RLHF

Data Collection

Freelancer Overview

TESLA — AI Resident Intern (Autopilot), multi-modal data fusion and lane-detection data pipeline work (Summer 2023).. Brings 3+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Internal, Proprietary Tooling, and AWS SageMaker. Education includes Master of Science, Stanford University (2024) and Bachelor of Science, University of Texas at Austin (2022). AI-training focus includes data types such as 3D Sensor, Computer Code, and Programming and labeling workflows including Data Collection, Evaluation, and Rating.

Labeling Experience

LLM Distiller (Data Curation & Alignment) — open-source distillation and alignment project.

OtherTextFine-tuning

Developed an open-source LLM distillation workflow focused on high-fidelity dataset curation and fine-tuning alignment to preserve benchmark performance. Distilled large 70B-parameter models down to 7B parameters while maintaining evaluation quality within a small performance margin. Curated and aligned training data to support downstream model behavior targets. • Curated datasets for high-fidelity distillation. • Performed fine-tuning alignment to preserve benchmark performance. • Built an LLM distiller pipeline for 70B→7B compression. • Validated performance retention within an estimated 3% margin.

2022 - 2024

TESLA — AI Resident Intern (Autopilot), multi-modal data fusion and lane-detection data pipeline work (Summer 2023).

3D SensorData Collection

Worked on multi-modal data fusion by collaborating on how to fuse and annotate camera and radar streams for automated driving spatial awareness. Developed and applied targeted dataset filtering as part of a high-precision computer-vision data pipeline to improve detection quality in challenging conditions. Focused on creating evaluation-ready training data from real sensor inputs. • Annotated and prepared multi-modal camera/radar data for a multi-modal transformer. • Built lane detection pipelines with targeted dataset filtering. • Reduced false positives in low-light conditions through data/pipeline optimization. • Partnered with an Autopilot team to curate training inputs for model performance improvements.

2023 - 2023

AMAZON WEB SERVICES (AWS) — Software Engineering Intern (ML Platforms), training-data monitoring and drift detection (Summer 2022).

AWS SageMaker

Engineered scalable data ingestion and monitoring capabilities to support real-time model monitoring and training data curation at high event throughput. Integrated automated performance drift detection into deployment pipelines to flag when live data diverged from training distributions. Ensured curated datasets remained aligned with model training inputs through continuous evaluation. • Ingested and monitored 500k+ events/sec for model training data curation. • Added drift detection metrics to detect live-vs-training distribution divergence. • Supported ongoing dataset quality checks through automated monitoring. • Contributed to maintaining evaluation-ready training data for ML platforms.

2022 - 2022

UT AUSTIN AI LAB — Undergraduate Research Assistant, RL reward curation and exploration research (2021–2022).

RLHF

Conducted research into reinforcement learning reward curation for autonomous navigation, focusing on optimizing reward functions and data density in sparse-reward settings. Worked on data-centric RL problem formulation to shape learning signals used for training. Contributed to understanding how curated RL rewards affect exploration efficiency in complex environments. • Researched RL algorithms with emphasis on reward-function optimization. • Improved data density strategies for sparse reward environments. • Supported autonomous navigation training via curated reward signals. • Co-authored ICML 2022 work on efficient exploration strategies in complex AI environments.

2021 - 2022

AI Resident Intern (Autopilot) — lane detection computer vision pipeline development

ImageSegmentation

Developed and optimized computer vision pipelines for lane detection as part of an Autopilot AI resident role, improving detection quality in low-light conditions. This work involved generating and using training data labels/annotations implicitly to train and evaluate the vision model. Performance improvements were validated through reduced false-positive detections during inference.

2023

Show more

Education

Work History