- Currently working on PyPI Packages for Data Science, including optimrl for GRPO based RL training of language models
- Currently learning how to train LLMs from scratch with open source stacks: tokenizer training, pretraining, supervised fine-tuning, and RL
- Exploring fine-tuning open source models (LoRA, QLoRA, TRL) and LLM evaluation, inference, and serving, with experiment tracking on MLflow
- Interested in AI agents, tool use, and retrieval augmented generation
- Currently learning efficient training of small open source language models on consumer hardware using Unsloth, Axolotl, and MLX, with GGUF quantization, llama.cpp, Ollama, and LM Studio for local inference
- Ask me about Machine Learning, NLP, LLMs, and Data Science
- Check out my Kaggle Profile for competitions and notebooks
optimrl Group Relative Policy Optimization for Efficient RL Training |
outlier-cleaner A Python package for detecting and removing outliers in data using various statistical methods and advanced distribution analysis |
| Programming Languages |
|
| Machine Learning & Data Science |
|
| LLM & AI |
|
| Web & App Frameworks |
|
| Databases |
|
| Cloud & DevOps |
|
| Visualization & BI |
|
| Other Tools |
|
|
|
|
|
|
|
- 🎉 Merged PR #9 in SubaashNair/subashanan-nair-portfolio