Foundational models & architectures
Transformers, diffusion, multimodal systems, and the mechanics underneath.
Machine learning researcher · engineering leader
Making AI
understand
the world.
I work at the intersection of foundational models, content representation, and responsible AI — turning hard research questions into systems people can use.
Atlanta, GA / currently building
at Harman International
at Harman International
01 / Curiosity map
Areas I explore
A set of connected questions around how machines represent, generate, recommend, and make decisions about content.
Content representation
Making audio, visual, and text content legible to machines.
Disentanglement
Separating signal from spurious correlation in feature spaces.
Responsible AI
Fairness, safety, privacy, robustness, and evaluation.
Generated content detection
Understanding synthetic media and building reliable detectors.
Recommendation systems
Personalized discovery powered by meaningful representations.
Computer vision
From pixels and events to robust perception in the wild.
02 / In practice
Where I’ve worked
A career moving from image formation and perception to foundation models, safety, and product-scale ML.
2024 — now
Senior Principal Engineer, Machine Learning
Harman International
Leading multimodal content representation and recommendation systems from research direction through data architecture, model development, and product integration.
2022 — 2024
Senior Machine Learning Engineer / Researcher
Meta Platforms, Inc.
Built responsible AI systems, synthetic content detection, fairness mitigations, and safety solutions for multimodal generative products.
2018 — 2021
Computer Vision & Deep Learning Researcher
Alibaba DAMO Academy · Israel Lab
Created event recognition and automatic photo album curation systems, including a 50% improvement in event detection and segmentation F1.
2005 — 2018
Computer Vision Algorithms Developer
Elbit Systems · Rafael · Israeli Air Force
Developed perception algorithms for detection, recognition, registration, change detection, and multi-target tracking.
03 / The record
Selected research
My Google Scholar profile ↗ is the living index. A few recent threads are below.
2025
Fine-Grained Erasure in Text-to-Image Diffusion-based Foundation Models
K. Thakral, T. Glaser, T. Hassner, M. Vatsa, R. Singh
2024
The Llama 3 Herd of Models
A. Dubey et al., including T. Glaser
2024
Navigating Text-To-Image Generative Bias Across Indic Languages
S. Mittal, A. Sudan, R. Singh, M. Vatsa, T. Glaser, T. Hassner
2024
On responsible machine learning datasets emphasizing fairness, privacy and regulatory norms
S. Mittal et al., including T. Glaser
2022
PETA: Photo Albums Event Recognition using Transformers Attention
T. Glaser, E. Ben-Baruch, G. Sharir et al.
04 / Foundations
Built on strong fundamentals
Technion — Israel Institute of Technology
M.Sc. Electrical Engineering
Honors · GPA 4.0 / 95.4 · Computer vision research thesis
Technion — Israel Institute of Technology
B.Sc. Electrical Engineering
& B.A. Physics
Dual degree · Psagot excellence track · Honors