
Itai Gafni
AI & cyber founder; ex-Unit 8200 data scientist.
Igafni
Tel Aviv-Yafo, Israel
Joined November 2025
Network
15.1K connectionsRWONOH
OOSHDS
IRMSST
TBAMLO
SHYMGK
YBAK
MBGBTO
Summary
Founder focused on AI and cyber security: Itai has founded and led multiple early-stage startups (Softwareye, Huskeys, and a stealth company) building AI-driven products for monitoring and web/API security. These ventures show a pattern of shipping data‑driven products that combine ML research techniques with practical security and observability tooling. startuphub+2
Practitioner and technical writer in applied ML/monitoring: He publishes technical, code‑heavy articles (e.g., DQN+ARIMA for resource scaling; SmartShot for zero‑shot classification with NLI) demonstrating hands‑on experience with ML architectures, time‑series forecasting, and productionization considerations. medium+1
Background in Unit 8200 — technical and cyber R&D experience: Public professional references and the person's own posts identify them as an ex‑8200 engineer/scientist who worked across data engineering, data research, cyber R&D and later AI/data science lead responsibilities — a common pathway into Israeli deep‑tech and cyber startups. linkedin+1
Active in Israeli startup networks and accelerators: participation in programs and networks (e.g., Ignite DeepTech cohort references and connections to 8200 alumni/angel networks) indicates engagement with accelerators, investor networks, and the Israeli deep‑tech ecosystem. startuphub+2
Work
Education
Projects
Writing
SmartShot: Fine‑Tuning Zero Shot Classification Models with NLI
November 29, 2023A practical guide on fine‑tuning zero‑shot classification/NLI models (MoritzLaurer/mDeBERTa etc.), dataset preparation, and training considerations — targeted at applied ML practitioners.
The Power of AI in Monitoring: Leveraging DQN & ARIMA for Optimizing Resources Scaling
September 8, 2023A technical article describing combining Deep Q‑Networks (DQN) with ARIMA time‑series forecasting to optimize resource scaling in multi‑cloud monitoring contexts; includes implementation notes and example code (notebook snippets).