Dhruv Madeka
Research Engineer, Google
Biography
Dhruv Madeka is a Research Engineer at Google DeepMind working on Gemini pre-training. His current research focuses on Large Language Modeling. Before that, Dhruv was a Principal Machine Learning Scientist at Amazon working on applying Deep Reinforcement Learning to inventory management problems. Dhruv has also worked on developing generative and supervised deep learning models for probabilistic time series forecasting. In the past - Dhruv worked in the Quantitative Research team at Bloomberg LP, developing open source tools for the Jupyter Notebook and conducting advanced mathematical research in derivatives pricing, quantitative finance and forecasting.
Interests
- Large Language Models
- Natural Language Processing
- Reinforcement Learning
- Forecasting
Recent Publications
(2023).
Contextual Bandits for Evaluating and Improving Inventory Control Policies.
arXiv preprint arXiv:2310.16096.
(2023).
Learning an Inventory Control Policy with General Inventory Arrival Dynamics.
arXiv preprint arXiv:2310.17168.
(2023).
Scaling Laws for Imitation Learning in NetHack.
arXiv preprint arXiv:2307.09423.
(2023).
Linear Reinforcement Learning with Ball Structure Action Space.
ALT 2023.
(2022).
A few expert queries suffices for sample-efficient rl with resets and linear value approximation.
NeurIPS ‘22’.
(2022).
Deep Inventory Management.
arXiv preprint arXiv:2210.03137.
(2022).
MQRetNN: Multi-Horizon Time Series Forecasting with Retrieval Augmentation.
Milets 2022.
(2021).
A Framework for the Meta-Analysis of Randomized Experiments with Applications to Heavy-Tailed Response Data.
arXiv preprint arXiv:2112.07602.
(2020).
Mqtransformer: Multi-horizon forecasts with context dependent and feedback-aware attention.
Milets 2022.
(2019).
All roads lead to quantitative finance.
Quantitative Finance.
Contact
- madeka@bu.edu
- New York, NY