Skip to main content
Kavita KumariKK
Looking for a job

Kavita Kumari

@kavitakumari1

AI security researcher at Technical University of Darmstadt building defenses and detectors for deepfakes and model misalignment.

Germany
Message

At Technical University of Darmstadt - System Security Lab, I design machine-learning models for security-critical and multimodal data, and defenses against backdoors, data leakage, and adversarial behavior. I also build evaluation pipelines for LLM safety and robustness.

I designed VoiceRadar, a system for detecting AI-generated and manipulated speech. Its paper received the Distinguished Paper Award at NDSS 2025, and the system is being commercialized with an industry partner.

For Light2Lie, I developed a physics-inspired vision pipeline that uses reflectance, illumination, and signal inconsistencies to distinguish AI-generated images from real ones. I evaluated its robustness under distribution shifts and adversarial perturbations.

In my current AI Security Research Fellowship, I’m testing whether a reasoning model’s hidden context can be recovered from its output logits. I also built a LoRA fine-tuning pipeline for studying self-narrative language drift and emergent misalignment.

Experience

Work history, roles, and key accomplishments

HF
Current

AI Security Research Fellow

Heron AI Security Research Fellowship

Sep 2026 - Present (2 months)

Testing whether a reasoning model's hidden context (system prompt or chain of thought) can be recovered from its output logits, a channel that refusal training and output filtering do not cover.

Technical University of Darmstadt logoTD
Current

Applied Postdoctoral Researcher

Technical University of Darmstadt

Dec 2022 - Present (3 years 11 months)

Designed and implemented physics-informed and energy-based ML models for security-critical and multimodal data analysis. Designed defenses for privacy and security threats in distributed learning setups and LLM evaluation pipelines for model safety and robustness.

PythonPyTorch
Scientific Machine Learning
MathematicsComputer VisionAI Safety
Physics Informed Modeling
Signal Processing
Bayesian Modeling
Explainable AIOptimizationModel Evaluation
Repeatability
LLMGenerative AI
Multimodal Analysis
Speech Analytics
OpenAI APIAI GuardrailsLoRAs
Misalignment Issues
University of Texas at San Antonio logoUA

Teaching / Research Assistant (PhD Student)

University of Texas at San Antonio

Aug 2016 - Aug 2022 (6 years)

Developed machine learning and probabilistic modeling solutions for privacy, security, and uncertainty analysis in real-world systems. Designed game-theoretic and stochastic models to analyze adversarial behavior and repeated interactions in data-driven systems.

PythonPyTorch
Probabilistic Modeling
Game Theory
Stochastic Processes
Dynamical Systems
Experimental DesignResearchMachine LearningExplainable AI
AI ML Security
Volatility
Bayesian Modeling
Federated Learning

Education

Degrees, certifications, and relevant coursework

Aligarh Muslim University logoAU

Aligarh Muslim University

Grade: 8.9 / 10.0

Pursued studies at Aligarh Muslim University, achieving a GPA of 8.9 out of 10.0.

Tech stack

Software and tools used professionally

Availability

Looking for a job

Location

Germany

Authorized to work in

Salary expectations

90k+ EUR

Job categories

Skills

AI SecurityMachine Learning
LLM Safety
Deepfake Detection
Federated Learning
Adversarial Machine Learning
Privacy
Bayesian Machine Learning
Bayesian Modeling
Computer VisionSignal Processing
Physics Informed Modeling
Explainable AIGame Theory
Stochastic Processes

Interested in hiring Kavita?

You can contact Kavita and 600K+ other talented remote workers on Himalayas.

Message Kavita

People also viewed

View all talent

Get matched with your dream remote job

Sign up now and join over 1,000,000+ remote workers who receive personalized job alerts, curated job matches, and more for free!

Sign up
Himalayas profile for an example user named Frankie Sullivan