Ora Nova Fandina
Ora Nova Fandina

Ora Nova Fandina

אורה פנדינה

I am an AI Research Scientist with strong foundations in the theory of machine learning. I work on understanding and improving AI systems through rigorous evaluation and precise analytical reasoning.

My research interests include reliable evaluation of generative AI and the reliability of automated evaluators. I am also interested in computational efficiency in AI, building on my work in dimensionality reduction and metric embeddings.

I hold a Ph.D. in Computer Science from the Hebrew University of Jerusalem, where I was advised by Yair Bartal and Ofer Neiman, and completed postdoctoral research at Aarhus University with Kasper Green Larsen.

Publications

2026

Beyond Blind Spots: Analytic Hints for Mitigating LLM-Based Evaluation Pitfalls
AAAI 2026 Workshop · Paper
Ora Nova Fandina, Eitan Farchi, Shmulik Froimovich, Raviv Gal, Wesam Ibraheem, Rami Katan, Alice Podolsky

2025

Vintage Code, Modern Judges: Meta-Validation in Low Data Regimes
ASE 2025 Workshop · Paper
Ora Nova Fandina, Gal Amram, Eitan Farchi, Shmulik Froimovich, Raviv Gal, Wesam Ibraheem, Rami Katan, Alice Podolsky, Orna Raz
Automated Validation of LLM-based Evaluators for Software Engineering Artifacts
Preprint, 2025 · Paper
Ora Nova Fandina, Eitan Farchi, Shmulik Froimovich, Rami Katan, Alice Podolsky, Orna Raz, Avi Ziv
Exploring Straightforward Methods for Automatic Conversational Red-Teaming
NAACL 2025 · Industry Track · Paper
George Kour, Naama Zwerdling, Marcel Zalmanovici, Ateret Anaby-Tavor, Ora Nova Fandina, Eitan Farchi

2024

How Safe is Your Safety Metric? Automatic Concatenation Tests for Metric Reliability
Preprint, 2024 · revised 2025 · Paper
Ora Nova Fandina, Leshem Choshen, Eitan Farchi, George Kour, Yotam Perlitz, Orna Raz

2023

Unveiling Safety Vulnerabilities of Large Language Models, GEM Workshop at EMNLP 2023 [arXiv]
George Kour, Marcel Zalmanovici, Naama Zwerdling, Esther Goldbraich, Ora Nova Fandina, Ateret Anaby-Tavor, Orna Raz, Eitan Farchi
The Fast Johnson-Lindenstrauss Transform is Even Faster, ICML 2023 [arXiv]
with Mikael Møller Høgsgaard and Kasper Green Larsen
Haiku abstract
An old method's pace // revised, now faster to use. // Efficiency reigns.
Barriers for Faster Dimensionality Reduction, STACS 2023 [arXiv]
with Mikael Møller Høgsgaard and Kasper Green Larsen

2022

Optimality of the Johnson-Lindenstrauss Dimensionality Reduction for Practical Measures, SoCG 2022 [arXiv] [slides] [video]
with Yair Bartal and Kasper Green Larsen
Haiku abstract
It was not designed // to preserve the average. // Yet, it optimally does.

2020

Online Probabilistic Metric Embedding: A General Framework for Bypassing Inherent Bounds, SODA 2020 [pdf]
with Yair Bartal and Seeun William Umboh
Haiku abstract
Online random trees. // Tight. Yet, the error is large. // Are they still of use?

2019

Dimensionality Reduction: Theoretical Perspective On Practical Measures, NeurIPS 2019 [pdf] [full version] [slides] [poster] [code] [video]
with Yair Bartal and Ofer Neiman
Haiku abstract
They say, in practice, // few dimensions are enough. // But in theory?
Covering Metric Spaces by Few Trees, ICALP 2019 [arXiv] [slides]
with Yair Bartal and Ofer Neiman
Haiku abstract
In a doubling world, // You see this grove of trees? // They are enough.

2015

Holographic parallel processor for calculating Kronecker product, Natural Computing 14 (2015), 433-436 [pdf]
with Shlomi Dolev and Joseph Rosen

2013

Succinct Permanent is NEXP-hard with Many Hard Instances, CIAC13 [pdf]
with Shlomi Dolev and Dan Gutfreund

Patents & applications

Modifying Artificial Neural Networks for Testing Security
US20260050673A1 · Published patent application · February 19, 2026 · Paper
Ora Nova Fandina, Orna Raz, George Kour, Marcel Zalmanovici, Eitan Daniel Farchi, Ateret Anaby-Tavor

Technical writing

Talks & outreach

ML Concepts Seminar

A study seminar at IBM Research in Haifa, with Eitan Farchi and Ora Fandina, exploring connections between probability, statistics, decision theory, game theory, optimization, and machine learning.

Watch the seminar ↗

Teaching

  • Metric Embedding Theory and its Algorithmic Applications, at HUJI. Lecturer, Fall 2016 - Fall 2019 [Lecture notes]

  • Data Structures, at BGU. Lecturer, Fall 2014 - Fall 2016

  • Algorithms, at Israeli Flight Academy. Lecturer, Fall 2013 - Fall 2015

Personal

I was born and grew up in Ukraine, where I finished high school. I arrived in Israel together with my family in my teens.

Dan is my husband and Ruth, Esther, and Sarah are our daughters. Together with Dan we have backpacked in:

Ethiopia, Georgia, Tajikistan, Armenia, Azerbaijan, Kyrgyzstan, Japan, Costa-Rica, Spain, Netherlands, Germany, Greece, Italy, Slovakia, Egypt, Turkey, Jordan.

The Hebrew language

The Academy of the Hebrew Language is where you can propose new Hebrew words and read about the history of the Hebrew language.

Math / CS reading list
  • Kurt Godel’s Letter to John von Neumann (1956), [pdf]
  • More is Different (1972), P.W. Anderson [pdf]
  • Cargo Cult Science (1974), Richard Feynman [pdf]
  • Big omicron and big omega and big theta (1976), D.E. Knuth [pdf]
  • What Every Computer Scientist Should Know About Floating-Point Arithmetic (1991), D. Goldberg [pdf]
  • All I really need to know...(1993), D. P. Stern [pdf]
  • Possible Trends in Mathematics in the Coming Decades (1998), M. Gromov [pdf]
  • The Vinous Shock: How to Open a Bottle with a Book (2018), M. Lewi [pdf]
  • Ideas That Created the Future: Classic Papers of Computer Science (2021), H.L. Lewis
My books on Goodreads ↗