Curriculum Vitae

Download PDF

Research Scientist and Applied Scientist with 10+ years in machine learning, NLP, and information retrieval. I build, deploy, and evaluate production LLM and agentic AI systems, with a Ph.D. from UMass Amherst and 20+ peer-reviewed publications.

Experience

• Research Scientist, Dataminr — Aug. 2022 – present
  • Lead production AI initiatives end to end — from problem definition through deployment, monitoring, and iteration — shipping generative and agentic LLM systems that process 200M+ real-time events per day.
  • Design and deploy domain-specialized LLMs and multi-agent agentic AI systems that autonomously reason over high-volume real-time data streams to surface critical events, threats, and risks.
  • Build evaluation systems and tooling to measure the grounding, reliability, calibration, and robustness of LLM and agentic outputs in production; apply fine-tuning, knowledge distillation, and quantization to ship custom, efficient models.
  • Active industry publication record (ACL 2025, NAACL 2024, CIKM 2023, SIGIR 2023); mentor junior researchers and contribute to Dataminr’s AI for Good program.
• Postdoctoral Researcher, Brown University (Prof. Carsten Eickhoff) — Dec. 2020 – Jul. 2022
  • Led an independent research agenda on uncertainty-aware and fairness-constrained ranking; mentored Ph.D. and undergraduate researchers and delivered all grant milestones on schedule.
• Research Assistant, UMass Amherst (Prof. W. Bruce Croft) — Sep. 2015 – Dec. 2020
  • Developed novel training methods and robustness techniques for neural IR and QA models, yielding publications at SIGIR, EMNLP, ICTIR, ICML, and ECIR and two best-paper awards.
• Site Project Manager, UMass Amherst (IARPA MATERIAL) — Sep. 2017 – Dec. 2020
  • Led UMass Amherst deliverables for a multi-university IARPA grant; coordinated biweekly reporting and annual cross-lingual IR system evaluations, achieving top retrieval scores across all competing teams.
• Research Intern, Microsoft Research (Dr. Fernando Diaz, Dr. Bhaskar Mitra) — Feb. 2019 – Jun. 2019
  • Applied dense retrieval to build a scalable policy index for task transfer in reinforcement learning, enabling efficient policy lookup without expensive pairwise comparisons.
• Research Intern, Microsoft Research (Dr. Katja Hofmann, Dr. Bhaskar Mitra) — May 2017 – Aug. 2017
  • Developed an adversarial domain adaptation method for neural ranking models that generalizes to out-of-distribution markets with minimal labeled data (SIGIR 2018 Best Short Paper).

Education

• Ph.D. in Computer Science, University of Massachusetts Amherst, 2020
• M.S. in Computer Science, University of Massachusetts Amherst, 2017
• B.A. in Computer Science and Mathematics, New York University, 2015

Technical Skills

• Programming: Python, PyTorch, TensorFlow, JAX, CUDA, SQL, Bash, Java, C
• AI & LLMs: Large Language Models, Agentic AI & Multi-Agent Systems, Retrieval-Augmented Generation (RAG), Generative AI, Fine-Tuning, Knowledge Distillation, Quantization, RLHF, Neural Ranking, Information Retrieval, NLP
• Evaluation & Reliability: LLM & Agentic Evaluation (grounding, reliability, actionability), Calibration, Uncertainty Quantification, Robustness under Distribution Shift, Fairness & Bias Mitigation, Online Learning
• Production & Systems: End-to-end deployment, High-throughput real-time streaming, Performance monitoring, Latency/throughput thresholds, Model evaluation pipelines
• Frameworks & Tools: Hugging Face Transformers, Scikit-learn, NumPy, SciPy, Pandas, Galago, PyLucene, Linux/Unix, git

Selected Publications

A full, up-to-date list is on my publications page and Google Scholar.

• Explain then Rank: Scale Calibration of Neural Rankers Using Natural Language Explanations from LLMs. P. Yu, D. Cohen, H. Lamba, J. Tetreault, A. Jaimes. ACL (Findings), 2025.
• In-Context Example Ordering Guided by Label Distributions. Z. Xu, D. Cohen, B. Wang, V. Srikumar. NAACL (Findings), 2024.
• Predictive Uncertainty-based Bias Mitigation in Ranking. M. Hauss, D. Cohen, M. Mansoury, M. de Rijke, C. Eickhoff. CIKM, 2023.
• A Lightweight Constrained Generation Alternative for Query-Focused Summarization. Z. Xu, D. Cohen. SIGIR, 2023.
• Learning a Better Negative Sampling Policy with Deep Neural Networks for Search. D. Cohen, S. Jordan, W.B. Croft. ICTIR, 2019. Best Full Paper Award.
• Cross Domain Regularization for Neural Ranking Models Using Adversarial Learning. D. Cohen, B. Mitra, K. Hofmann, W.B. Croft. SIGIR, 2018. Best Short Paper Award.

Awards & Grants

• Best Full Paper, ICTIR, 2019
• Best Short Paper, SIGIR, 2018
• IARPA BETTER & MATERIAL Research Grants, co-writer, 2017 & 2020
• Bloomberg Data Science Research Grant, co-writer, 2018

Service

• Program Committee / Reviewer: SIGIR, EMNLP, EACL, CIKM, WWW, WSDM, ACL, KDD, AAAI, TOIS, TACL