Orchid Majumder

I am a senior staff software engineer at Google Workspace, working on long-horizon agentic evaluation and agentic recursive self-improvement for Gemini Agent. Previously, I was a principal research engineer at Amazon AGI focused on LLM pre-training, where I worked on the NeMo-Megatron based distributed training stack used to train the Amazon Nova models on thousands of AI accelerators on AWS. My interests there spanned distributed training, ML systems, architecture exploration and pre-training scaling laws.

Earlier, at AWS AI Labs, I worked on large-scale diffusion models, and before that on few-shot learning, meta-learning and hyperparameter optimization for visual recognition (see my publications below). I hold an MS from Columbia University and a BS from Jadavpur University, both in computer science.


Selected Publications

  • The Amazon Nova Family of Models: Technical Report and Model Card (arXiv) — Amazon AGI. [paper]
  • Efficient Scaling of Diffusion Transformers for Text-to-Image Generation (arXiv) — Hao Li, Shamit Lal, Zhiheng Li, Yusheng Xie, Ying Wang, Yang Zou, Orchid Majumder, R. Manmatha, Zhuowen Tu, Stefano Ermon, Stefano Soatto, Ashwin Swaminathan. [paper]
  • On the Scalability of Diffusion-Based Text-to-Image Generation (CVPR 2024) — Hao Li, Yang Zou, Ying Wang, Orchid Majumder, Yusheng Xie, R Manmatha, Ashwin Swaminathan, Zhuowen Tu, Stefano Ermon, Stefano Soatto. [paper]
  • Supervised Momentum Contrastive Learning for Few-Shot Classification (arXiv) — Orchid Majumder, Avinash Ravichandran, Subhransu Maji, Alessandro Achille, Marzia Polito, Stefano Soatto. [paper]
  • Incremental Few-Shot Meta-Learning via Indirect Discriminant Alignment (ECCV 2020) — Qing Liu, Orchid Majumder, Alessandro Achille, Avinash Ravichandran, Rahul Bhotika, Stefano Soatto. [paper]
  • MARTHE: Scheduling the Learning Rate Via Online Hypergradients (IJCAI 2020) — Michele Donini, Luca Franceschi, Orchid Majumder, Massimiliano Pontil, Paolo Frasconi. [paper]
  • d-SNE: Domain Adaptation Using Stochastic Neighborhood Embedding (CVPR 2019 Oral) — Xiang Xu, Xiong Zhou, Ragav Venkatesan, Gurumurthy Swaminathan, Orchid Majumder. [paper]

GitHub

  • AdaTune — Adaptive learning rate tuning for PyTorch