Chandan Singh csinva

Chandan Singh

AI & interpretability researcher · Staff Research Scientist at Voyage AI, training large embedding models
_{Previously ~4 years at Microsoft Research (Deep Learning group) · PhD in AI at UC Berkeley with Bin Yu}

🔍 interpretability · 💊 clinical decision rules · 🧠 semantic brain mapping · 🤖 LLMs

🔍 Interpretability

imodels Interpretable, accurate, sklearn-compatible predictive modeling JOSS '21 FIGS·PNAS '22 HSTree·ICML '22 Agentic‑imodels·NeurIPS '26
imodelsX Interpretability for text and LLMs Aug‑imodels·Nat. Comms '23 Tree‑Prompt·EMNLP '23 iPrompt·ICLR wkshp '23 SASC·NeurIPS wkshp '23 QA‑Embs·NeurIPS '24
deep-explanation-penalization Penalize neural-network explanations to align models ICML '20
hierarchical-dnn-interpretations Hierarchical interpretations for neural network predictions ICLR '19
adaptive-wavelets Replace neural nets with adaptive, interpretable wavelet models NeurIPS '21

💊 Clinical decision rules

covid19-severity-prediction Extensive COVID-19 data + forecasting for counties and hospitals HDSR '21
clinical-self-verification Improving LLM-based clinical information extraction via self-verification ML4H '23
clinical-rule-development Building and vetting clinical decision rules with data and LLMs IAI CDI·PLOS DH '22 Patient perspectives·Nat. SR '25 CDI bias·npj Dig. Med. '25 HACHI·npj Dig. Med. '26 BC‑LLM·NeurIPS '25

🧠 Semantic brain mapping

automated-brain-explanations Building natural-language explanations of fMRI language responses GCT·Nat. Neuro '26 QA models·bioRxiv '25
interpretable-embeddings Encoding models built from LLM answers to questions NeurIPS '24

🤝 Selected projects advised

token-economizer Experience-driven black-box cost control for coding agents (led by Aaron Li) arXiv '26
sim-dino Simplifying DINO via coding rate regularization (led by Ziyang Wu) ICML '25
pasta Post-hoc attention steering for LLMs (led by Qingru Zhang) ICLR '24
meta-tree Learning a decision tree algorithm with transformers (led by Yufan Zhuang) TMLR '24
student-sim Training LLM-based student simulators for testing AI tutors (led by Ke Yang) arXiv '26
mixture-of-inputs Text generation beyond discrete token sampling (led by Yufan Zhuang) NeurIPS '25

🧰 Notes & tools

ml-ai notes Notes and resources on AI
vflow Utilities for trustworthy data science JOSS '21

🌐 Open-source contributions

autogluon Fast and accurate AutoML in a few lines of code
big-bench Collaborative benchmark for measuring and extrapolating the capabilities of language models
nl-augmenter Collaborative repository of natural-language transformations for data augmentation

Smaller contributions: interpretable-ml-book · conference-acceptance-rates · awesome-machine-learning-interpretability · awesome-llm-interpretability · awesome-interpretable-machine-learning · executable-books · deep-fMRI-dataset · iterative-random-forest

Earlier deep learning & ML projects

gan-vae-pretrained-pytorch Pretrained GANs + VAEs + classifiers for MNIST/CIFAR in PyTorch
gpt-paper-title-generator Generating paper titles with GPT-2
disentangled-attribution-curves Attribution curves for interpreting tree ensembles arXiv '19
matching-with-gans Matching in GAN latent space for better bias benchmarking CVPR wkshp '21
data-viz-utils Functions for easily making publication-quality figures with matplotlib
mdl-complexity Revisiting complexity and the bias-variance tradeoff JMLR '21