Tal Shapira

From 2009 to 2012, Shapira was a cadet in the Israel Defense Forces’ Talpiot excellence program, and in 2012 he received a bachelor’s degree in physics, with a minor in mathematics, from the Hebrew University of Jerusalem.⁠[1][1]

From 2013 to 2019, Shapira worked in Israel’s Office of the Prime Minister: as an RF and antennas engineer from 2013 to 2015, a data science researcher from 2015 to 2016, and a data science team leader from 2016 to 2018, during which he received the Israel Defense Award, and as head of a cybersecurity R&D branch from 2018 to 2019.⁠[1][1][1][1]

In 2018, Shapira received a master’s degree in electrical engineering from Tel Aviv University, graduating magna cum laude.⁠[1][2]

In 2019, Shapira and Tel Aviv University professor Yuval Shavitt presented the FlowPic method at an IEEE INFOCOM 2019 workshop: it transforms basic network flow data into a picture and then uses convolutional neural networks to identify the flow category (browsing, chat, video and so on) and the application in use, classifying even VPN and Tor traffic with high accuracy and identifying applications with 99.7% accuracy. The research was also shown that year at Cyber Week 2019 and AI Week 2019.⁠[3][3][3][3][3][4][4]

From 2019 to 2020, he was head of algorithms, covering deep learning and computer vision, at the heads-up display startup Guardian Optical Technologies, which was later acquired by Gentex.⁠[1][5][2]

In 2020, Shapira co-founded the cybersecurity company Reco with Ofer KleinOfer KleinOfer Klein is the co-founder and chief executive of the cybersecurity company Reco, having previously founded Kwik, an internet of things shopping button company; he has led Reco’s shift from enterprise collaboration and SaaS security to AI agent security.Open the full entry and Gal NakashGal NakashGal Nakash is an Israeli cybersecurity practitioner and the co-founder and chief product officer of the cybersecurity company Reco. Before joining Reco he led research at Israel’s Office of the Prime Minister; at Reco he is responsible for product and writes about AI agent security.Open the full entry, initially serving as co-founder and chief scientist.⁠[5][1]

The same year, he and Shavitt introduced BGP2VEC, which borrows from natural language processing models to treat BGP route announcements as sentences and embed each autonomous system number as a vector. At NOMS 2020 they used the method to infer the business relationships between autonomous systems with 95.2% accuracy; at the ACM SIGCOMM NetAI 2020 workshop they used it to detect BGP hijacks, detecting 32 of 48 past hijack events between 2008 and 2018.⁠[3][3][3][3][3][3]

In 2021, Shapira received a PhD in electrical engineering from Tel Aviv University, again graduating magna cum laude. That year an extended version of FlowPic was published in IEEE Transactions on Network and Service Management, and the SASA (Source-Aware Self-Attention) method was published in IEEE/ACM Transactions on Networking; the latter uses multiple geolocation services and their confidence to detect malicious routes and point out the most problematic router along the path.⁠[1][2][3][3][3]

On January 15, 2021, Shapira’s Tel Aviv University team (TAU-Team) won the NetML Network Traffic Analytics Challenge 2020, which asked participants to use machine learning to detect malicious flows or distinguish between applications at a fine-grained level; across its seven tracks, TAU-Team was the only team to place in the top three in every track and to win first place in more than one.⁠[6][6][6][6]

On October 6, 2021, he presented research on detecting IP hijack attacks with deep learning at Israeli Networking Day 2021, held at Ben Gurion University.⁠[4]

In 2022, he and Shavitt published BGP2Vec and AP2Vec, an unsupervised BGP hijacking detection method, in IEEE Transactions on Network and Service Management, and with co-authors he published a paper in Computer Communications on classifying encrypted traffic with neural ordinary differential equations. As of his personal website’s last update, he was a postdoctoral researcher at the School of Computer Science at Reichman University, took part in Red Hat Research’s “CVE mining and prediction” project, reviewed for journals including IEEE/ACM Transactions on Networking, and served on the technical program committee of IFIP Networking 2022.⁠[3][3][3][3][2][7][1][1]

When Reco announced additional funding in April 2025, Shapira lived in Tel Aviv according to his LinkedInLinkedInLinkedIn is a social networking platform for professionals. It was founded in 2002 by Reid Hoffman and others, launched in 2003, and listed on the New York Stock Exchange in 2011; after Microsoft acquired it for approximately $26.2 billion in 2016, it became a wholly owned Microsoft subsidiary. Its revenue comes from membership subscriptions, advertising, and recruitment solutions.Open the full entry profile.⁠[8]

As of September 2026, Shapira was Reco’s co-founder and chief technology officer and a member of the Cloud Security Alliance’s AI Controls Security Working Group. On September 24, 2026, he wrote an explainer on the company blog on the OWASP Top 10 for LLM Applications, arguing that prompt injection and excessive agency compound AI risk and that large language models cannot reliably separate trusted instructions from untrusted data.⁠[9][10][11][11]

Personal website: https://talshapira.github.io/ (opens in a new window)⁠[2]

Reco: Official website: https://www.reco.ai/ (opens in a new window).⁠[10]

X: Public account: https://x.com/talshapirala (opens in a new window)⁠[2]

LinkedIn: Public profile: https://www.linkedin.com/in/tal-shapira/ (opens in a new window)⁠[10]

Google Scholar: Public profile: https://scholar.google.com/citations?user=2VthnhIAAAAJ (opens in a new window)⁠[2]

ORCID: https://orcid.org/0000-0002-7151-0927 (opens in a new window)⁠[2]