Cody Blakeney
Head Of Research at Arcee AI
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolCody Blakeney is a Head of Research and PhD computer scientist with a decade of experience building and scaling large-scale ML data pipelines and model-training tooling. He led LLM data efforts at Databricks Mosaic Research—creating Spark-based preprocessing for multi-trillion-token corpora and publishing work on training-time data interventions—and now directs research at Arcee AI. Skilled across Python, C++, CUDA and production ML stacks, he has contributed to prominent open-source LLM tooling (mosaicml/llm-foundry), adding features like flash-attention scaling and robust preprocessing for fine-tuning. His background in neural net compression and energy-efficient inference for edge devices informs a pragmatic approach to model and data engineering. Based in Austin, he combines academic rigor with production-first engineering to drive reproducible, scalable LLM research.
10 years of coding experience
10 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Texas State University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (19)
transformers10
tokenize10
pytorch10
python10
datasets10
llm10
tokenizer10
deep-learning10
natural-language-processing10
huggingface-datasets10
neural-network10
nlp10
wandb9
flash9
multihead-attention9
Programming languages (6)
Github contributions (5)
mosaicml/llm-foundry
May 2023 - Jul 2024
LLM training code for Databricks foundation models
Role in this project:
ML Engineer
ML EngineerContributions:22 reviews, 18 PRs, 23 pushes in 1 year 3 months
Contributions summary:Cody contributed to the development and improvement of the LLM training code within the repository. Their work includes adding features for custom group settings in a finetuning process using wandb, integrating softmax scaling into flash attention, and preparing datasets in JSON and CSV formats, along with associated preprocessing, tokenization and conversion scripts. The user also addressed bugs and made code formatting and refactoring changes to improve code quality and maintainability.
databricksfoundation-modelsllmdeep-learningneural-networks
codestar12/Parallel-Independent-Blockwise-Distillation
Jan 2020 - May 2021
Contributions:1 review, 172 commits, 2 PRs in 1 year 3 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.