→ 06.19.2026: New on arXiv: FAPO: Fully Automated Prompt Optimization of Multi-Step LLM Pipelines.

→ 05.22.2026: Self-Improving In-Context Learning is up on arXiv.

→ 05.13.2026: Named Best Reviewer at ICML 2026.

→ 04.30.2026: Test-Time Detoxification Without Training or Learning Anything accepted to ICML 2026.

→ 04.28.2026: Posted to arXiv: Test-Time Safety Alignment.

→ 02.03.2026: Test-Time Detoxification Without Training or Learning Anything—out as a preprint.

→ 12.18.2025: Our reasoning RAG model is recognized by Yejin Choi, Dawn Song, and others in their recent paper, Adaptation of Agentic AI, as a leading approach for agent adaptation with tool execution.

→ 11.10.2025: New preprint: Think Before You Retrieve: Learning Test-Time Adaptive Search with Small Language Models.

→ 10.25.2025: Large Language Models Encode Semantics and Alignment in Linearly Separable Representations headed to the IJCNLP and the Asian Chapter of ACL 2025 main conference.

→ 09.18.2025: Into NeurIPS 2025: Risk-Averse Constrained Reinforcement Learning with Optimized Certainty Equivalents.

→ 05.16.2025: Learning Task Representations from In-Context Learning in at ACL 2025 Findings.

→ 03.01.2025: Robust Intelligence acquired by Cisco! I’m now part of the founding team at Foundation AI.

→ 09.17.2024: Mitigating Off-Policy Bias in Actor-Critic Methods with One-Step Q-learning to appear in Transactions on Machine Learning Research.

→ 09.11.2024: Joined Robust Intelligence as a research intern!

→ 09.02.2024: Compatible Gradient Approximations for Actor-Critic Algorithms—now on arXiv.

→ 05.28.2024: Summer research intern at Google NYC!

→ 03.02.2024: Parameter-free Reduction of the Estimation Bias in Deep Reinforcement Learning for Deterministic Policy Gradients to appear in Neural Processing Letters.

→ 11.16.2023: Actor-Prioritized Experience Replay accepted to Journal of Artificial Intelligence Research.

→ 10.26.2023: Into Machine Learning: Deep Intrinsically Motivated Exploration in Continuous Control.

→ 10.23.2023: Deep Reinforcement Learning Based Joint Downlink Beamforming and RIS Configuration headed to the IEEE International Conference on Communications 2023 Workshop on Data Driven Intelligence.

→ 03.20.2023: Accepted offer from Yale University—starting Ph.D. in Electrical and Computer Engineering, Fall 2023.

→ 03.21.2022: Awarded the 5G and Beyond Fellowship from Türk Telekom.

→ 09.10.2021: Three papers accepted to IEEE ICTAI 2021—a start to the master’s.