- AI researcher, builder, innovator, educator, and writer turning AI ideas into teams, systems, products, and publications.
- Over 30 years of industry experience, including roles as Head of AI Research, VP AI, and Head of Data Science; built and hired research, data science, and engineering teams; led cross-functional AI work.
Leadership & Research Experience
Faculty Member - HIT, School of Computer Science
2024-Present
Conducts applied AI research and supervises student projects; teaches graduate AI courses at Bar-Ilan on LLMs, GenAI, and agents; created undergraduate Hands-On AI courses in LLMs and GenAI; served as Head of the Software Engineering track in 2024-2025.
Head of Data Science - Questar / SafeRide Technologies
2020-2024
Built and hired the data science team; led big-data AI for vehicle telematics across signal, audio, graph, anomaly detection, and onboard-model work; helped define a no-code product for onboard monitoring models.
VP of Artificial Intelligence - harmon.ie
2019-2020
Led cross-functional AI work for document management, including language processing and recommendation systems, connecting research direction with product and software execution.
Head of AI Research - Motorola Solutions, Israel Innovation Center
2015-2019
Built and led the research team at Motorola Solutions' first AI lab in the Innovation Center; designed the roadmap across video, text, speech, and public-safety AI.
AVP Data Science - Citi Israel Innovation Lab
2013-2015
Worked on financial time-series data science, anomaly detection, and trading-flow analytics.
Founder / CTO / R&D and algorithm leadership - SearchNG, CloudTuner, MPCode, Tegrity, AppiVision, Amdocs, Comverse, earlier roles
1991-2012
Founded ventures and led hands-on intelligent-system work in search, multimedia, video streaming, wireless communications, telecom, predictive modeling, handwriting recognition, imaging, inspection, and signal acquisition.
Research Themes
Pragmatic AI: methods for inference, generation, and coordination when clean benchmark assumptions do not hold.
Robust AI
Vision, language, and signals under degraded or ambiguous inputs.
Spec-driven GenAI
Generation of materials, text, images, and signals constrained by verifiable targets.
Multi-agent AI
Evaluation and coordination without trusted benchmarks or shared state.
Research, Patents & Systems
Publications
40+ Scholar-indexed works and 470+ citations across ML, computer vision, NLP, deep learning, and signal processing.
Patents
10 patent records across vehicle diagnostics, driver monitoring, model configuration, onboard inference, public-safety audio, and search/image matching.
Systems
Hands-on building of working systems in model integration, agent automation, runtime tracing, analytics, diffusion steering, evaluation, and reasoning platforms.