Rejith Joseph
Principal Engineer at Annapurna Labs
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolRejith Joseph is a Principal Engineer based in Seattle with a decade of experience building high-performance AI and cloud systems at Amazon and AWS. He designs and optimizes GPU- and ASIC-accelerated infrastructure for large language models and has driven production ML work spanning search, autocomplete, spell correction, and LLM optimization. His background combines hands-on systems engineering—HPC, EMR, ECS, and distributed GPU orchestration—with machine learning contributions, including benchmarks and training code for Amazon’s open-source DSSTNE deep learning engine. Known for taking prototypes to customer demos and leading small technical teams, he blends low-level performance tuning with product-focused delivery. Rejith’s early research produced GPU-optimized algorithms presented at top conferences, reflecting a sustained interest in squeezing performance from modern hardware. He brings a practical mix of research rigor and operational experience that helps move cutting-edge ML from lab to scale.
10 years of coding experience
20 years of employment as a software developer
Bachelors, Computer Science, Bachelors, Computer Science at National Institute of Technology CalicutMasters, Computer Science, Masters, Computer Science at University of Florida
Github Skills (7)
neural-network10
machine-learning10
deep-learning10
tensorflow10
python10
autoencoder10
benchmark9
Programming languages (2)
Github contributions (5)
amazon-archives/amazon-dsstne
May 2016 - Feb 2020
Deep Scalable Sparse Tensor Network Engine (DSSTNE) is an Amazon developed library for building Deep Learning (DL) machine learning (ML) models
Role in this project:
ML Engineer
ML EngineerContributions:52 commits, 20 PRs, 34 pushes in 3 years 10 months
Contributions summary:Rejith contributed to the project by adding and modifying benchmarks related to a sparse autoencoder implemented in TensorFlow. Their commits show the implementation of a feedforward neural network, data loading, and training loops. They updated training modes and made adjustments to core training code, demonstrating an understanding of the model's architecture and training process.
deep-learningmachine-learningtensor
rgeorgej/fairseq
Aug 2021 - Aug 2021
Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
Contributions:2 pushes in 1 day
nlpsequencepythonmachine-learningfacebook
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.