Aravind Srinivas is an Indian-American computer scientist, co-founder and chief executive of Perplexity, and a former research scientist at OpenAI and Google DeepMind. He completed his PhD in computer science at the University of California, Berkeley in 2021 under the supervision of Pieter Abbeel and Trevor Darrell. He co-founded Perplexity in August 2022 with Denis Yarats, Andy Konwinski, and Johnny Ho, and has remained chief executive through the company's growth from a four-person prototype into one of the most valuable consumer-AI search products of the 2024 to 2026 cycle.
At a glance
- Education: Bachelor of Technology in electrical engineering, Indian Institute of Technology Madras (2017); PhD in computer science, UC Berkeley (2021).
- Current role: Co-founder and Chief Executive Officer of Perplexity since August 2022.
- Notable prior affiliations: Research intern at OpenAI (2019, 2020, 2021); research intern at Google DeepMind (2018); research scientist at OpenAI (2021 to 2022).
- Key research contributions: Co-author of "Decision Transformer: Reinforcement Learning via Sequence Modeling" (NeurIPS 2021); co-author of papers on contrastive self-supervised learning at Berkeley including "CURL: Contrastive Unsupervised Representations for Reinforcement Learning" (ICML 2020) and "Reinforcement Learning with Augmented Data" (NeurIPS 2020).
- X / Twitter: @AravSrinivas
- LinkedIn: aravind-srinivas-16051987
Origins
Srinivas was born in 1994 in Chennai, India. He completed his secondary education in Chennai and entered the Indian Institute of Technology Madras in 2013, where he pursued a Bachelor of Technology in electrical engineering with a focus on signal processing and machine learning. During his undergraduate years he developed an interest in deep learning, which was reaching mainstream academic attention through the rise of convolutional neural networks and the success of AlexNet, AlphaGo, and the early sequence-to-sequence models. He completed his bachelor's degree in 2017.
He moved to the United States in 2017 to begin a PhD in computer science at UC Berkeley under the supervision of Pieter Abbeel, the deep-reinforcement-learning researcher and OpenAI co-founder, and Trevor Darrell, the computer-vision researcher and Berkeley Artificial Intelligence Research Lab co-director. The doctoral research focused on self-supervised representation learning and reinforcement learning, two of the most actively explored areas of deep learning during the late-2010s scaling era. He defended in 2021.
Career
The Berkeley PhD overlapped with three summer research internships at OpenAI (2019, 2020, 2021) and one at Google DeepMind (2018). The Decision Transformer paper, which framed reinforcement learning as a conditional sequence-modeling problem in the same way GPT framed language modeling, came out of the 2021 OpenAI internship period and was co-authored with Lili Chen, Kevin Lu, Igor Mordatch, and Pieter Abbeel. It became one of the more-cited reinforcement-learning papers of the early-2020s offline-RL wave.
After defending his PhD in 2021, Srinivas joined OpenAI as a research scientist. The tenure was short. By mid-2022 he had departed to start a company focused on a problem he believed was specifically suited to the new generation of large language models: search and information retrieval. He co-founded Perplexity in August 2022 with Denis Yarats (a former Meta AI research scientist), Andy Konwinski (a Databricks co-founder), and Johnny Ho (a Quora engineer). The company launched a public preview of its conversational answer engine in December 2022.
The Perplexity thesis was that the search-engine experience had not been redesigned around large language models in a serious way, and that a system which combined live retrieval with strong language-model summarisation could compete with general-web-search incumbents on a meaningful subset of query types. The product positioning was not "a chatbot" or "a research tool" but "a search engine," competing directly with Google and Bing rather than with ChatGPT or Claude. The framing turned out to be unusually effective at fundraising and at recruiting a category-creation narrative in the press.
The company raised a $25.6 million Series A in March 2023 led by NEA, a $73.6 million Series B in January 2024 led by IVP at a $520 million valuation, and a Series C of approximately $250 million in April 2024 at a $1 billion valuation that made Perplexity one of the fastest paths to unicorn status in the post-2022 AI cycle. Subsequent rounds in 2024 and 2025 reportedly priced the company at $9 billion (June 2024) and at $18 billion (November 2025), with strategic investments from Nvidia, Jeff Bezos, and others. The capital-side trajectory follows the talent-leads-capital pattern documented in the diaspora map: Srinivas's frontier-lab credentials (DeepMind, OpenAI) compressed the time-to-credibility for institutional investors in a way that a non-frontier-lab founder could not have achieved at the same speed.
Affiliations
- Indian Institute of Technology Madras: Bachelor of Technology student, 2013 to 2017.
- UC Berkeley: PhD student, 2017 to 2021.
- Google DeepMind: Research intern, summer 2018.
- OpenAI: Research intern (2019, 2020, 2021), then research scientist (2021 to 2022).
- Perplexity: Co-founder and Chief Executive Officer, August 2022 to present.
Notable contributions
- Decision Transformer: Reinforcement Learning via Sequence Modeling (June 2021). Co-author of the paper that reframed offline reinforcement learning as a conditional sequence-modeling problem, applying the transformer architecture to RL trajectories. Cited extensively in the subsequent offline-RL literature and the early-2020s foundation-model-for-RL discussion.
- CURL: Contrastive Unsupervised Representations for Reinforcement Learning (April 2020). Co-author of the paper applying contrastive self-supervised learning to RL state representations, presented at ICML 2020.
- Reinforcement Learning with Augmented Data (April 2020). Co-author of the RAD paper showing data augmentation as a simple, broadly applicable strategy for sample-efficient pixel-based reinforcement learning, presented at NeurIPS 2020.
- Perplexity co-founding (August 2022). Co-founder and CEO of the conversational answer-engine startup, which by mid-2026 reports tens of millions of monthly active users and partnerships with browser and device makers for default-search placement.
Open questions
- Perplexity's competitive position against general-web-search incumbents. Google launched AI Overviews and AI Mode in 2024 and 2025, putting LLM-summarised answers directly into the dominant search interface. Whether Perplexity can sustain growth on the strength of differentiated UX, source-citation defaults, and a faster product iteration cycle, or whether the general-web-search incumbent's distribution advantage closes the gap, is the central question for the company's 2026 to 2027 trajectory.
- The browser play. Perplexity launched a custom browser (Comet) in 2025 and has been reported to have explored acquiring or building deeper distribution surfaces. Srinivas has publicly framed the browser as a strategic priority. Whether the browser channel produces meaningful user share or remains a small slice of the distribution mix is unclear.
- Capital intensity vs revenue trajectory. The 2025 round at an $18 billion valuation prices Perplexity at multiples that imply substantial revenue growth ahead. Public revenue disclosures have been limited; the gap between the valuation and reported revenue is large enough that the next round will need to show a clear traction inflection.
- Frontier-model dependence. Perplexity relies on third-party frontier models from OpenAI, Anthropic, and increasingly its own Sonar fine-tunes for some surfaces. Whether the company moves toward training its own frontier-scale models, continues as a routing-and-orchestration layer over third-party models, or pursues some hybrid will shape its cost structure and its strategic optionality.
Sources
- Aravind Srinivas Google Scholar profile. Publication record, citation counts, and the timeline of academic output from the Berkeley PhD period.
- Decision Transformer: Reinforcement Learning via Sequence Modeling. The June 2021 paper that defined the offline-RL-as-sequence-modeling framing.
- Perplexity raises $73M to advance its AI search. TechCrunch coverage of the Series B at a $520 million valuation.
- Perplexity, the AI search startup, hits $1B valuation. Reuters coverage of the April 2024 Series C round.
- How Perplexity wants to take on Google Search. Verge interview with Srinivas on the search-incumbent thesis.
- Aravind Srinivas profile in Forbes 30 Under 30. Background biography and early career trajectory.
- The Atlas people-layer extracted 2026-05-11, which records Srinivas's distinct affiliations at Google DeepMind, OpenAI, and Perplexity.
- Companion essay: The diaspora map for the broader pattern of frontier-lab-to-consumer-product career arcs that Srinivas's trajectory anchors.
