Scholar
Rohan Pandey
Google Scholar ID: j7OhJCEAAAAJ
OpenAI
Citations & Impact
Citations
296
H-index
6
i10-index
5
Publications
11
Co-authors
7
list available
Contact
Publications
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
- Selected publications: 'gzip Predicts Data-dependent Scaling Laws' (ArXiv 2024); 'Multimodal Learning Without Multimodal Data: Guarantees and Applications' (ICLR 2024); 'Towards Vision-Language Mechanistic Interpretability: a Causal Tracing Tool for BLIP' (ICCV 2023 - CLVL); 'Cross-modal Attention Congruence Regularization for Vision-Language Relation Alignment' (ACL 2023); 'Syntax-guided Neural Module Distillation to Probe Compositionality in Sentence Embeddings' (EACL 2023); 'A Family of Cognitively Realistic Parsing Environments for Deep Reinforcement Learning' (NeurIPS 2021 - Deep RL).
Research Experience
- Explored model architecture questions spanning pre-training, reinforcement learning, and inference at OpenAI; built a multimodal web agent generating 5k lines of code weekly at Reworkd (YC S23); fine-tuned language models to automate enterprise-scale data annotation at Microsoft AI.
Education
- Graduated from Carnegie Mellon University in 2023, with an honors thesis on semantics in multimodal LLMs.
Background
- Research interests include teaching machines to do science, and has worked on training GPT-5 class models at OpenAI. Graduated from Carnegie Mellon University in 2023 with an honors thesis on semantics in multimodal LLMs.
Miscellany
- Runs a biweekly Sanskrit reading group in San Francisco; worked on OCR for Sanskrit to immortalize the classical Indian literary canon in the training corpus for superintelligence; forked ved/acc from e/acc in 2023; lived at AGI House SF, a hacker house in Twin Peaks, until September 2024; taught a Classical Indian Philosophy course at Carnegie Mellon University; conlanging in middle school led him to linguistics, and consequently to NLP & Sanskrit.
