Shamane Siri
Research Scientist at Pluralis Research
Melbourne, Victoria, Australia
Summary
🤩
Rockstar🎓
Top SchoolShamane Siri is a Research Scientist based in Melbourne with nine years of experience specializing in post-training and scalable alignment of foundation models across centralized and decentralized settings. He holds a PhD in Foundation Model Adaptation and a Master’s in Deep Reinforcement Learning, and has led applied NLP teams to productionize customized LLMs ranging from 7B to 70B parameters. His work spans agentic reasoning, reinforcement-tuning of LLM agents, and multimodal VLLM research for inclusive education, bridging rigorous research with deployable systems. At Oracle and Arcee.ai he designed agentic workflows and adaptive RLHF strategies that improved long-horizon decision-making and cost-efficient inference via Test Time Scale optimization. An active open-source contributor, he has refined RAG implementations in the widely used Hugging Face transformers repo, surfacing practical fixes for PyTorch-Lightning and distributed retrievers. Shamane’s profile mixes deep academic foundations with hands-on engineering for enterprise-scale model customization and robustness in real-world pipelines.
9 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy - PhD, Natural Language Processing, Doctor of Philosophy - PhD, Natural Language Processing at The University of Auckland
English
Stackoverflow
4,021reputation
565kreached
43answers
61questions
jupyter-notebook
top-5%
tensorflow
top-5%
keras
top-5%
nlp
top-5%
deep-learning
top-5%
Github Skills (22)
pytorch10
python10
machine-learning10
pytorch-lightning10
deeplearning-ai10
retrieval-augmented-generation10
deep-learning10
transformer10
nlp10
retrieval-augmented10
language-model9
tensorflow9
keras9
ray9
jupyter-notebook9
Programming languages (9)
Github contributions (5)
huggingface/transformers
Jun 2021 - Jun 2022
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:
ML Engineer
ML EngineerContributions:15 reviews, 5 commits, 8 PRs in 1 year
Contributions summary:Shamane's contributions primarily focused on refining and updating the RAG (Retrieval-Augmented Generation) model implementation within the `transformers` repository. They addressed compatibility issues with the latest PyTorch-Lightning (PL) version, including adapting the training framework, and incorporated updates related to Ray distributed retriever. The user also made minor fixes and refactored code related to the RAG end-to-end pipeline, suggesting an active role in improving its functionality and integration within the broader project.
pythonbertspeech-recognitionstate-of-the-artflax
shamanez/BERT-like-is-All-You-Need
Jun 2020 - Feb 2021
Contributions:18 commits, 6 pushes, 49 comments in 8 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.