Ivan Zhang

Ivan Zhang is a Canadian software engineer and entrepreneur. He is a co-founder of Cohere, the Toronto-headquartered enterprise-AI company he established in 2019 with Aidan Gomez and Nick Frosst, and was the company's chief technology officer through the founding period before transitioning into a senior leadership capacity in 2024. As of May 2026, he serves as a co-founder in senior leadership at Cohere following the April 24, 2026 announcement of a merger with the German AI company Aleph Alpha, backed by a Schwarz Group commitment and a Series E that values the combined entity at approximately $20 billion.

At a glance

• Education: Computer science at the University of Toronto, attended without completing the undergraduate degree; entered industry as a software engineer.
• Current role: Co-founder of Cohere since September 2019, in a senior leadership capacity following the chief-technology-officer period through 2024. Phil Blunsom holds the chief-technology-officer title as of mid-2025.
• Key contributions: co-founder of Cohere; co-founder of FOR.AI (later Cohere For AI) in 2017; co-author of Targeted Dropout at the NeurIPS 2018 CDNNRIA workshop with Geoffrey Hinton; co-author of Unsupervised Cipher Cracking Using Discrete GANs at ICLR 2018; co-author of Mitigating Harm in Language Models with Conditional-Likelihood Filtration (2021).
• Recognition: The Logic readers' innovation leader of 2024 alongside co-founders Gomez and Frosst; Maclean's AI Power List 2023 entry with the same co-founders.
• X / Twitter: @1vnzh
• LinkedIn: 1vnzh
• GitHub: 1vn
• Personal site: ivanzhang.ca

Origins

Zhang grew up in Toronto. He has described himself in public bios as a Chinese immigrant and a Torontonian, and has cited a high-school computer-science class as the entry point into programming. He enrolled at the University of Toronto for a computer-science undergraduate degree but left before completing it, deciding that hands-on engineering work would teach him faster than the rest of the program. The University of Toronto Department of Computer Science later identified him in alumni coverage as a "former computer science student" rather than a graduate, distinguishing him from his Cohere co-founders Gomez and Frosst, who completed their U of T undergraduate degrees.

After leaving university, Zhang took an early-career role at a friend's startup in Toronto as a back-end and infrastructure engineer, and went on to work as a software engineer in the biotech and machine-learning operations space; the specific employers have not been disclosed in public press materials. He began collaborating with Aidan Gomez during this period, while Gomez was an undergraduate at U of T who would join Google Brain as a research intern in 2017 and co-author the Attention Is All You Need paper that introduced the Transformer architecture.

Career

In 2017, Zhang and Gomez founded FOR.AI, an independent collaborative machine-learning research program. The program was structured as a network of "friends, classmates, and engineers" working on machine-learning research outside any single institution, with the explicit aim of broadening the pipeline of researchers from non-traditional backgrounds. FOR.AI later became the basis for Cohere's research arm and was renamed Cohere For AI in 2022 under Sara Hooker.

Zhang's published research from this period sits in the Hinton orbit at U of T. He is a co-author of Targeted Dropout, the NeurIPS 2018 CDNNRIA workshop paper that introduced a self-reinforcing-sparsity dropout method for training prunable neural networks; co-authors include Aidan Gomez, Kevin Swersky, Yarin Gal, and Geoffrey Hinton. He is also a co-author of Unsupervised Cipher Cracking Using Discrete GANs at ICLR 2018, with first author Aidan Gomez and co-authors including Lukasz Kaiser, one of the eight Transformer-paper authors. The two papers established his presence at the boundary of the U of T and Google Brain Toronto research communities before Cohere existed.

In 2019, Zhang and Gomez decided to start a company. Press coverage from the period frames the decision around two observations: first, the rapid uptake of Transformer-based architectures inside Google after the 2017 paper, which Gomez had observed during his Google Brain internship, and second, the February 2019 release of GPT-2, which suggested that frontier language-model capability could be a commercializable enterprise infrastructure layer. Nick Frosst, then a researcher at Google Brain Toronto under Geoffrey Hinton, joined as the third co-founder. Cohere was incorporated in Toronto in September 2019.

Zhang took the chief-technology-officer title at the founding and the operational responsibility for the engineering work that turned research models into a production API. The first commercial Command model launched in November 2021, after Cohere closed its $40 million Series A in September 2021 led by Index Ventures, with Radical Ventures and the AI researchers Geoffrey Hinton, Fei-Fei Li, Pieter Abbeel, and Raquel Urtasun participating. Subsequent generations included Command R+ (April 2024), Command R (May 2024), Command R7B (December 2024), and Command A (March 2025), with the Embed and Rerank retrieval-augmented-generation product line released alongside. Zhang is listed as an author on the Command A technical paper (March 2025) and on the Aya Expanse paper (December 2024) from Cohere For AI, which formally launched in 2022 under Sara Hooker.

Through 2024 and 2025, Zhang transitioned out of the chief-technology-officer title into a senior leadership capacity. The transition was not the subject of a public announcement; the Inovia Capital founder profile from this period describes him as having moved from a foundational software-engineering role to leading "Cohere's product organization, focusing on product management and the developer experience." Saurabh Baji subsequently held the chief-technology-officer title before departing in 2025, and the Cohere About page lists Phil Blunsom, the former DeepMind and Oxford researcher, as Cohere's chief technology officer as of mid-2025. The April 24, 2026 Aleph Alpha merger and accompanying Series E commitment from the Schwarz Group, the German parent of Lidl and Kaufland, transitions Cohere into a transatlantic sovereign-AI consolidation entity at a reported approximately $20 billion combined-entity valuation. Zhang continues in his senior co-founder role at the combined company.

Affiliations

• Friend's startup (Toronto): Back-end and infrastructure software engineer, mid-2010s. Specific employer not publicly disclosed.
• Software-engineering work in biotech and MLOps: pre-Cohere employer history, late 2010s. Specific employers not publicly disclosed.
• FOR.AI: Co-founder with Aidan Gomez, 2017. Independent collaborative machine-learning research program; later folded into Cohere For AI.
• Cohere: Co-founder, September 2019 to present. Chief Technology Officer through the founding period, transitioning to senior leadership in 2024.
• Cohere For AI: Co-founder, 2022 to present (with Aidan Gomez and Nick Frosst, building on the FOR.AI foundation; Sara Hooker leads day-to-day operations).

Notable contributions

Zhang's published record is concentrated in the late-2010s U of T and Toronto-Google-Brain research-community window, with subsequent author credits on Cohere production-research papers. His Google Scholar profile lists approximately 546 citations and an h-index of 11 as of mid-2026.

• Targeted Dropout (NeurIPS 2018 CDNNRIA workshop). Co-author with Aidan Gomez, Kevin Swersky, Yarin Gal, and Geoffrey Hinton on the paper introducing a self-reinforcing-sparsity dropout method for training prunable networks. The most-cited paper in his record at over 140 citations.
• Unsupervised Cipher Cracking Using Discrete GANs (ICLR 2018). Co-author with Aidan Gomez, Sicong Huang, Bryan M. Li, Muhammad Osama, and Lukasz Kaiser; first attention-equipped generative-adversarial-network attack on substitution ciphers.
• Mitigating Harm in Language Models with Conditional-Likelihood Filtration (August 2021). Co-author with Helen Ngo, Cooper Raterink, João G. M. Araújo, Carol Chen, Adrien Morisot, and Nick Frosst, applying a conditional-likelihood filtering approach to safety-relevant pretraining-data selection.
• Predicting Twitter Engagement with Deep Language Models (RecSys 2020 Challenge). Co-author with the Maksims Volkovs team on the engagement-prediction track.
• FOR.AI co-founding (2017). Co-founder with Aidan Gomez of the independent ML research network that preceded Cohere For AI.
• Cohere co-founding (September 2019). Co-founder and original chief technology officer.
• Command production-research papers (2024 to 2025). Author credits include the Aya Expanse paper (December 2024) and the Command A technical paper (March 2025).
• Public communications. Long-form podcast appearance on the Madrona Founded & Funded episode (August 2023) covering retrieval-augmented generation, foundation-model deployment for enterprise, and feedback loops; episode 101 of the All Quiet on the Second Front podcast (September 2025) covering air-gapped AI deployment in defense, healthcare, and finance.
• Recognition. Maclean's AI Power List 2023 entry alongside Gomez and Frosst; The Logic readers' innovation leader of 2024 with the same co-founders.

Investments and boards

• Cohere (AI): Co-founder, September 2019 to present. Privately held. Cumulative private capital of approximately $1.6 billion across Series A through Series C extension pre-merger, with a Series E in process at a reported $20 billion combined-entity post-merger valuation anchored by the Schwarz Group.
• Cohere For AI (AI): Co-founder, 2022 to present. Non-profit open-research division operated within Cohere's research-and-development budget.
• Borderless AI (Software / AI): Angel investor, January 2025. Personal participation alongside Aidan Gomez in a $5 million seed extension that brought Borderless AI's total seed funding to $32 million; Borderless AI's HR-agent platform is built on Cohere's models.

No public personal angel-investor activity is on record in semiconductors, datacenters, or energy as of May 2026.

Network

Zhang's longest-running professional relationships are with his Cohere co-founders Aidan Gomez (chief executive officer) and Nick Frosst, who together have been the founding triad since September 2019 and the FOR.AI collaboration since 2017. Gomez has been the closest of the early collaborators: the Targeted Dropout, Unsupervised Cipher Cracking, and FOR.AI lines all sit on the Zhang-Gomez axis, and the decision to start Cohere followed several years of joint research output. Frosst was Google Brain Toronto's first dedicated employee under Geoffrey Hinton, the deep-learning researcher and 2024 Nobel Laureate in Physics, and a co-author on the original capsule-network papers; the Hinton connection was the academic bridge that drew Hinton and his Toronto venture firm Radical Ventures into Cohere's investor base.

The Targeted Dropout co-authors anchor most of his other publicly visible academic relationships. Geoffrey Hinton, Yarin Gal at the University of Oxford, and Kevin Swersky at Google Brain Toronto were the senior researchers on that line, and Lukasz Kaiser, the Transformer-paper co-author now at OpenAI, was the senior collaborator on Unsupervised Cipher Cracking. Helen Ngo, the first author on the conditional-likelihood-filtration paper, was a Cohere research collaborator during the 2021 period. Among broader frontier-research peers, Zhang has been the lower-profile of the Cohere founders by public-facing appearance count and has not been a regular fixture on the Davos, WSJ Tech Live, or TIME 100 AI circuit that Gomez has anchored.

Position in the field

As of May 2026, Zhang is one of three co-founders of the only Canadian-headquartered frontier-capability AI lab and the most engineering- and product-focused of the founding triad. Press coverage has consistently identified the three co-founders as filling complementary roles: Gomez as chief executive officer and the public face on enterprise positioning and sovereign-AI deals, Frosst as the open-research and Cohere For AI lead, and Zhang as the engineering and product systems builder who carried the early infrastructure and currently leads the developer-experience and product organization.

The transition from chief-technology-officer to senior leadership in 2024, with the title subsequently held by Saurabh Baji and then Phil Blunsom, is the most visible operational change in his profile. Industry coverage has characterized the move as a re-allocation of the chief-technology-officer title to a non-co-founder executive rather than a departure from technical responsibility, with Zhang continuing in product and developer-experience leadership.

The non-traditional educational pedigree, with no completed undergraduate degree, is the second structurally distinctive element of his profile. Within the global pool of frontier-AI-lab co-founders, the self-taught pathway is rare. Zhang has used public appearances to advocate for hiring builders from non-traditional backgrounds and has connected this directly to the FOR.AI founding mandate.

Outlook

Open questions over the next 6 to 18 months:

• Aleph Alpha merger integration. Operational integration through 2026 and 2027, with implications for the product-organization rationalization Zhang oversees on the Cohere side.
• Series E close. Whether Cohere's Series E closes at the reported $20 billion combined-entity valuation and which additional strategic investors join the Schwarz Group.
• Product organization scope. The forward scope of Zhang's product-and-developer-experience role as the merged company integrates the Aleph Alpha PhariaAI platform and the Berlin-Toronto operating-anchor structure.
• Public-profile cadence. Whether the merger-period spotlight expands Zhang's media presence beyond engineering-track venues, or the co-founder distribution of public-facing work remains as it has been.
• Personal angel activity. Whether the Borderless AI investment foreshadows a broader pattern of personal angel activity in the Cohere-customer ecosystem.

Sources

• Ivan Zhang on Inovia Capital. Founder profile from a Cohere investor describing the foundational software-engineering origin and the current product-organization leadership role.
• Ivan Zhang on SALT speakers. Speaker bio from the SALT iConnections New York 2023 conference covering the FOR.AI co-founding, biotech and MLOps software-engineering experience, and Toronto background.
• Ivan Zhang Google Scholar profile. Citation metrics, h-index, and chronological publication record.
• Ivan Zhang personal site. Personal links page including GitHub, Google Scholar, and a curated publication list.
• Targeted Dropout. The 2019 arXiv release of the NeurIPS 2018 CDNNRIA workshop paper introducing targeted dropout, with Gomez, Zhang, Swersky, Gal, and Hinton as co-authors.
• Unsupervised Cipher Cracking Using Discrete GANs. The ICLR 2018 paper with Gomez as first author and Zhang as a co-author with Lukasz Kaiser.
• Mitigating Harm in Language Models with Conditional-Likelihood Filtration. The 2021 Cohere paper on safety-relevant pretraining data selection, with Helen Ngo as first author and Zhang as a co-author.
• Cohere's Ivan Zhang on Foundation Models, Retrieval Augmented Generation, and Feedback Loops. The August 2023 Madrona Founded & Funded podcast episode covering Zhang's RAG-versus-fine-tuning framework and Cohere's enterprise positioning.
• 101. Ivan Zhang, Co-Founder of Cohere | All Quiet on the Second Front. The September 2025 podcast episode covering air-gapped AI deployment in defense, healthcare, and finance.
• Cohere co-founders are The Logic readers' innovation leaders of 2024. The Logic 2024 reader-survey recognition of the three Cohere co-founders.
• Maclean's: The Power List on Ivan Zhang, Aidan Gomez, and Nick Frosst. The 2023 Maclean's Power List entry naming all three co-founders.
• Cohere About. Current leadership listing identifying Phil Blunsom as chief technology officer and Zhang as a co-founder.
• Cohere Raises $40 Million in Series A Financing. The September 2021 GlobeNewswire announcement of the Index Ventures-led Series A.
• Borderless AI Launches HRGPT With Strategic Funding from Cohere. The January 2025 announcement noting Aidan Gomez and Ivan Zhang as personal participants in the $5 million seed extension.
• Cohere AI Chronicle. Turing Post long-form chronicle covering the founding-period collaboration between Zhang and Gomez and the GPT-2-era decision to start Cohere.