Vicki Cheung

Vicki Cheung is a software engineer and startup founder known for her work building infrastructure for some of the most influential technology companies of the past decade. She was an early engineer at Duolingo, a founding engineer and Head of Infrastructure at OpenAI, a Senior Staff Software Engineer at Lyft, and is now the co-founder and CTO of Gantry, a machine learning observability and evaluation platform. Her work has significantly influenced MLOps practices, bridging research and production systems for large-scale AI and language-learning platforms.

Cheung's career spans the full stack, from building Duolingo's Android app and web frontend to designing distributed training systems for cutting-edge AI research at OpenAI. She has been a key figure in developing the engineering practices now known as MLOps, helping to bridge the gap between research and production machine learning systems.

While she maintains a relatively private personal life, Cheung is a respected voice in the engineering community, frequently speaking about SRE, MLOps, and lessons learned from scaling infrastructure at high-growth startups. She is known for her practical, hands-on approach and her ability to build reliable systems from the ground up. [edit]

Vicki Cheung was born in Hong Kong in the late 1980s or early 1990s (exact birthdate undisclosed). She grew up in Hong Kong, where she first encountered computers around age 10 in elementary school. Public school classes introduced her to Logo and required students to hand-code HTML websites using basic tools, as no advanced software was available. She found this empowering, realizing a 10-year-old could publish content online via services like GeoCities that masked her age. This sparked a lifelong passion for building digital things.

As a teenager, she self-taught coding by building PHP websites and web projects as hobbies, quickly viewing programming as both enjoyable and a viable career. In high school, she attended a summer camp at Carnegie Mellon University (CMU), took college-level classes, and decided to drop out of high school to enroll full-time at CMU around 2007. She graduated in 2011 after studying a computing-related field, most likely Computer Science. While at CMU, she joined Luis von Ahn's research group. This connection led directly to her first major opportunity: joining the project that became Duolingo.

Her college years and early work at Duolingo gave her experience in full-stack development, speech recognition, and building scalable products from scratch. This period was pivotal in shaping her career trajectory toward infrastructure engineering and machine learning systems. [edit]

Cheung's career is defined by her roles at high-growth technology companies. At Duolingo, she was an early engineer who helped the company grow from a team of four to forty. She built the Android app, much of the web frontend, and the API layer, and led development of the Incubator, a crowdsourcing platform that allowed Duolingo to rapidly expand its language offerings. She also contributed to speech recognition grading.

She then became engineering lead at TrueVault, a data privacy and compliance startup, where she worked on core platform APIs and infrastructure. But her most high-profile role came when she joined OpenAI as a founding engineer and Head of Infrastructure. At OpenAI, she designed and led the infrastructure for training large-scale machine learning models, including distributed training systems and cluster management, during the organization's early years. She is sometimes inaccurately described as a co-founder; she has clarified she was a founding engineer, not one of the official co-founders (Elon Musk, Sam Altman, Greg Brockman, Ilya Sutskever, etc.).

After OpenAI, Cheung became a Senior Staff Software Engineer at Lyft, focusing on SRE and MLOps. She then co-founded Gantry (with Josh Tobin and others), an ML infrastructure startup building tools for monitoring, evaluating, and improving machine learning systems in production. As CTO, she is responsible for the technical direction and platform infrastructure, helping companies operationalize their AI models. [edit]

Vicki Cheung projects a practical, hands-on, mission-driven engineering persona. Media and industry portray her as a respected, low-profile builder who bridges research and production in AI infrastructure. Her content (talks, interviews) is technical, pragmatic, and reflective, focusing on lessons from real scaling challenges rather than hype. There is no large public fandom, official fan name, or ritualized fan culture; her audience consists primarily of software engineers, MLOps practitioners, and AI researchers. She maintains a quiet, professional social media voice, with a modest presence on X/Twitter and LinkedIn, sharing insights on infrastructure and AI topics.[edit]

Vicki Cheung commands a unique kind of reverence within the technology sector, though her influence operates in the quiet corridors of engineering culture rather than the spotlight of pop stardom. There is no formal fan club, no specific moniker for her followers, and no organized campaigns or rituals dedicated to her name. Her audience consists almost entirely of software engineers, MLOps practitioners, and industry peers who revere her not for celebrity status but for her foundational contributions to large-scale AI infrastructure. The 'fandom' she cultivates is one of professional respect; her talks and interviews are consumed by those seeking pragmatic lessons on scaling systems, bridging the gap between research prototypes and production stability. This is a following built on technical utility and shared struggle against complex engineering challenges, devoid of the performative adulation seen in entertainment industries.

While her impact is profound among builders, iFANN has not yet gathered sufficient data to assign a global or category-specific fame ranking for Vicki Cheung. Consequently, no ordinal placement exists in the current iFANN database to indicate her standing relative to other public figures or within specific sectors like technology or engineering.[edit]

Vicki Cheung dropped out of high school after attending a Carnegie Mellon University summer camp to enroll in college early, showcasing her precocious drive for computer science. A native Cantonese speaker, she has volunteered with Bloomberg Startup and The Women's Foundation in Hong Kong to teach Python to young women, actively encouraging female participation in STEM fields. Her personal interests include EDM music, boxing, and organizing popular local board game meetups, reflecting a well-rounded life outside of code. Cheung has often shared how her early coding experiences in Hong Kong public schools were empowering because hand-writing HTML allowed her to publish content online while hiding her age from audiences. This early sense of digital anonymity fueled her confidence as a builder. She remains a featured speaker at major technical conferences, having discussed SRE best practices and MLOps challenges in interviews with HashiCorp and PowerToFly. Within the AI development community, a story persists about her pivotal role as a founding engineer at OpenAI, where she designed the distributed training systems that supported massive language models before departing to lead Gantry. Another notable chapter involves her time at Lyft, where she helped operationalize machine learning reliability, bridging the gap between research prototypes and production stability. These moments are less about viral spectacle and more about the foundational shifts she engineered that allowed large-scale AI to move from theory to reality.[edit]

Vicki Cheung keeps her family life private. There are no publicly available details about her parents, siblings, or extended family. She has not mentioned having a spouse or children in any of her professional talks or interviews. Her personal website mentions loving to build things while cuddling with her puppy, listening to EDM, organizing board game meetups, and taking up boxing. [edit]

Cheung is associated with several prominent technology companies and founders. She worked closely with Luis von Ahn at Duolingo, and with the founding team at OpenAI including Sam Altman and Greg Brockman. At Lyft, she collaborated with engineers across the company on reliability and ML infrastructure. She co-founded Gantry with Josh Tobin and others. [edit]

Vicki Cheung has not publicly disclosed any romantic relationships. There are no verified reports of partners, marriages, engagements, divorces, or children. She maintains a private personal life and does not discuss relationships in her professional talks or interviews. As of 2026, her relationship status is not publicly known. [edit]

There is no reliable public estimate of Vicki Cheung's net worth as of 2026. She is a private individual without publicly disclosed financial details. Her income sources include salaries and equity from her roles at Duolingo, TrueVault, OpenAI, and Lyft, as well as founder equity in Gantry. As a senior staff engineer at Lyft and a founding engineer at OpenAI, she likely earned compensation in the high six figures annually, including stock. However, without public filings or disclosures, any specific number would be speculative. [edit]