About
Biography: Parthasarathy (Partha) Ranganathan is currently part of the team at Google…
Activity
17K followers
-
Partha Ranganathan shared thisGemini Robotics 1.5 brings AI agents into the physical world at Google DeepMind's blog post at https://lnkd.in/gpC-5jXA Some really cool developments in robotics from Google.Partha Ranganathan shared this#Gemini Robotics 1.5 is a VLA that can now think 🧠 while taking action 🦾 , generating an internal sequence of reasoning using natural language 🤯. This makes the robot actions more interpretable (you can literally read the robot's thoughts), and unlocks more useful multi-step tasks. Learn more at https://lnkd.in/gkNmCieX Google DeepMind Robotics
-
Partha Ranganathan shared thisAnother interesting application of AI+science for weather prediction.Partha Ranganathan shared thisFor those of you interested in AI & weather prediction… While many discussions around AI focus on future potential benefits, one point that is sometimes missed is the scale at which AI is making a difference for people’s lives today. We’ve seen this with our work at Google in AI-enabled for weather forecasting, and advance notification and alerts for extreme weather events like floods (in over 100 countries) and wildfires (in over 27 countries), cyclones (see our Weather Lab) – and now we’re seeing impact at scale with advance weather predictions for monsoons. This summer, the University of Chicago used Google Research’s open-source NeuralGCM along with other advanced models to accurately predict the onset of the monsoon season in India up to a month in advance (including predicting an unusual dry spell). They worked with the Ministry of Agriculture & Farmers Welfare, Government of India to deliver advanced forecasts via SMS to 38 million farmers in India, allowing farmers to adjust planting decisions. Earlier University of Chicago research shows that an accurate forecast a month in advance gives farmers the tools to make informed decisions (e.g. planting timing, what type of crops, whether to buy more seeds, etc), and found these changes led to almost doubling their annual income. Exciting work and a strong example of the best of what AI and open source models can enable. You can read more about the project in our blogpost below: Or the coverage in the Wall Street Journal: https://lnkd.in/g4sKvvNw https://lnkd.in/ggq_SxeWHow AI is helping 38 million farmers with advance weather predictionsHow AI is helping 38 million farmers with advance weather predictions
-
Partha Ranganathan shared thisGoogle's contribution to OCP of our cool cooling work :)Partha Ranganathan shared thisGoogle has completed the contribution of its fifth-generation "Project Deschutes" Coolant Distribution Unit (CDU) specification to the Open Compute Project, a major milestone as the first such specification submission to the OCP Community! The project's primary goal is to accelerate the adoption of liquid cooling across the industry by providing a comprehensive, open design. The specification provides all the essential details, from technical design and manufacturing quality to maintenance and service procedures. It also includes the "Redmond" Hot Aisle Containment (HAC) data center structure, which is compatible with both CDUs and rear door heat exchangers. To help ensure broad deployment, the contribution outlines a supply chain strategy for worldwide availability, competitive pricing, and reduced lead times. See the 0.8 spec here: https://bit.ly/46iprnN
-
Partha Ranganathan shared thishttps://lnkd.in/gceMnj8E Nice blog from Praneet Singh Arshi and Joel Miller talking about how effective fleet management can be a great tool to manage our carbon footprint.Google’s approach to carbon-aware data center | Google Cloud BlogGoogle’s approach to carbon-aware data center | Google Cloud Blog
-
Partha Ranganathan shared thisOur work on AI for data science: we took a novel approach with a knowledge graph to for RAG-guided LLM reasoning coupled with a self-correcting agent to generate robust Python code. Read the paper for more details. Congrats to our intern Mossad Helali for his great work on this paper.Partha Ranganathan shared thisAlhamdulillah! Excited to announce our paper has been accepted at the EMNLP 2025 main conference! Automating Exploratory Data Analysis (EDA) is a crucial but challenging coding task. Current LLM-only methods can lack accuracy on private datasets and often depend on costly, proprietary models (like Gemini 2.5 Pro). Our solution? RAGvis, a novel RAG framework that uses a knowledge graph to guide LLM reasoning and a self-correcting agent to generate robust Python code. Key takeaways: - RAGvis achieves ~100% code pass rate & more than 2x semantic accuracy. - All with up to 60% less LLM tokens used. - RAGvis enables small, open-source LLMs (like Gemma 3) to achieve competitive performance. Huge thanks to our collaborators at Google and Kaggle and my supervisor! Yutai Luo, Tae Jun Ham, Jim Plotts, Ashwin Chaugule, Jichuan Chang, Partha Ranganathan, Essam Mansour Can't wait to present our work at EMNLP. We'll be open-sourcing our code and datasets soon!
-
Partha Ranganathan shared thisNano Banana! :) Google's viral image model aka Google 2.5 Flash Image is now available. I have had so much fun experimenting with it. If you haven't already, I would encourage you to take it for a drive!Partha Ranganathan shared this🍌 We have nano banana (aka Gemini 2.5 Flash Image) available on #VertexAI. Try now at https://lnkd.in/gPrxbsm6 This update brings state-of-the-art capabilities for creative use cases, allowing you to: - Fuse multiple images into a single, seamless visual. - Maintain character and style consistency across different generations. - Edit images conversationally with simple, natural language instructions. Developers and enterprises can now access Gemini 2.5 Flash Image in preview on Vertex AI, with built-in SynthID watermarking for responsible and transparent use. Read more about how customers like Adobe, Poe, and WPP are leveraging these new capabilities in our latest blogpost: https://lnkd.in/gnAzAj7Q
-
Partha Ranganathan reposted thisPartha Ranganathan reposted thisThere are a number of important questions on AI and energy that we’re working on at Google: How do we use AI to increase energy efficiency? How do we leverage AI to help address environmental and climate challenges? Can those approaches and their beneficial impact be scaled globally? At the same time, a key question is the amount of energy needed to use generative AI tools and apps, and if it can be reduced. To help answer this question, my colleagues and I did the math on the energy, carbon and water footprint per Gemini prompt. This is an area where state of the art is rapidly evolving, but we believe a clear understanding is important to enable us and others to continue to improve efficiency. Here’s a quick summary: We estimate the median Gemini App text prompt uses 0.24 watt-hours (Wh) of energy, emits 0.03 grams of carbon dioxide equivalent (gCO2e), and consumes 0.26 milliliters (or about five drops) of water. One way to explain the per prompt energy impact is that it's equivalent to watching TV for less than 9 seconds. We’re sharing our findings and methodology in our Technical Paper (link below) aiming to encourage industry-wide progress on more comprehensive reporting on the energy use of AI. We’ve been making rapid ongoing progress on the efficiency of our AI systems. Over a period of 12 months, as we continued to deliver higher-quality responses using more capable models, the median energy consumption and carbon footprint per Gemini Apps text prompt decreased by factors of 33x and 44x, respectively. All this thanks to our full stack approach to AI development, our latest data-center efficiencies, decades of experience serving software at scale (details in Technical Paper) – and the work of many teams at Google. To be clear, there is much more still to do. We’re continuing to improve efficiency at every layer of our operations (hardware, software, models, data centers), and we’re continuing to invest in energy infrastructure, grid improvements, and clean energy (existing tech and next-gen initiatives such as geothermal and nuclear). Thanks to my co-authors on this study Cooper Elsworth Tim Huang, Dave Patterson, Ian Schneider, Robert Sedivy, Savannah Goodman, Ben Townsend, Partha Ranganathan, Jeff Dean, Amin Vahdat, Ben Gomes. If you’d like to learn more, here are some resources: Blog post on our approach to energy innovation and AI: https://lnkd.in/gh8HX7zR Blog post on the math behind the numbers we shared today: https://lnkd.in/gQgq-Eir Technical Paper: https://lnkd.in/g8T88xMf
-
Partha Ranganathan reposted thisPartha Ranganathan reposted thisAs AI's role in society grows, so does the need for a deep understanding of its environmental impact. At Google, we're committed to not only advancing AI's capabilities, but also driving efficiency across the full stack of AI serving infrastructure. Driving improvements in AI efficiency requires comprehensive and accurate measurements to serve as baselines and to break down future opportunities. We found that existing calculations were either idealized or missed important quality optimizations. That is why we developed our own approach and applied it to our production AI serving stack. Alongside a remarkable team, @Jeff Dean and I are sharing a detailed methodology for measuring the energy, emissions, and water footprint of AI inference along with, to our knowledge, the first measurements of production AI serving. Published today, our technical paper estimates the median Gemini text prompt uses 0.24 watt-hours (Wh) of energy, emits 0.03 grams of carbon dioxide equivalent (gCO2e), and consumes 0.26 milliliters, or about five drops, of water. This per-prompt energy impact is equivalent to watching TV for less than nine seconds. What's more, our AI systems are getting significantly more efficient. Over a recent 12-month period, the energy and carbon footprint of a median Gemini text prompt dropped by 33x and 44x, respectively, as a result of efficiencies across Google’s entire stack. While we are proud of the innovation driving these efficiency gains, we are committed to the ongoing work needed to maximize performance per watt and continue such improvements in the years ahead. #Energy #Efficiency #Gemini #AI https://lnkd.in/gjpBecftMeasuring the environmental impact of AI inference | Google Cloud BlogMeasuring the environmental impact of AI inference | Google Cloud Blog
-
Partha Ranganathan shared thishttps://lnkd.in/g_inp4rK As the adage goes, with great power comes great responsibility. With great computing power comes great environmental responsibility as well, and so we've been very thoughtful about the environmental impact of the increasing computing demand caused by AI. Today, we discussed a detailed methodology for measuring the energy, emissions, and water footprint of AI inference along with, to our knowledge, the first measurements of production AI serving. Published today, our technical paper estimates the median Gemini text prompt uses 0.24 watt-hours (Wh) of energy, emits 0.03 grams of carbon dioxide equivalent (gCO2e), and consumes 0.26 milliliters, or about five drops, of water. This per-prompt energy impact is equivalent to watching TV for less than nine seconds. What's more, our AI systems are getting significantly more efficient. Over a recent 12-month period, the energy and carbon footprint of a median Gemini text prompt dropped by 33x and 44x, respectively, as a result of efficiencies across Google’s entire stack. Our technical paper at https://lnkd.in/gfNxxY4C goes into more details. As AI's role in society grows, so does the need for a deep understanding of its environmental impact. We're committed to not only advancing AI's capabilities, but also driving efficiency across the full stack of AI serving infrastructure. Also, check out our blog from Amin Vahdat and Jeff Dean at https://lnkd.in/gXtf5djr as well as more discussion from Ben Gomes at https://lnkd.in/g7g-xsEy. #Energy #Efficiency #Gemini #AIOur approach to energy innovation and AI’s environmental footprintOur approach to energy innovation and AI’s environmental footprint
Partha Ranganathan commented on a post
12h
Congrats, Manisha Gambhir
-
Partha Ranganathan liked thisPartha Ranganathan liked this🎉 Congratulations to Dr. Rahul Kande on successfully defending his Ph.D. at Texas A&M University! With 15 papers (and 2 under review) and a full-time role at AMD, Rahul’s work on hardware fuzzing has advanced the field and taught me a great deal along the way. I am proud of his perseverance, excited for him as he steps into the open field, and will certainly miss having him around the lab 🥳
-
Partha Ranganathan liked thisWhat a beautify story, Terry Morris! I too remember the first TechCon, when Dr. Frederick Kitson dropped on my table ~80 papers to review for the inaugural TechCon. He murmured something along the lines, "You are reviewing a lot for IEEE, these 80 papers will be nothing for you". Ever since, I have been hiding in every corner of Hewlett Packard Enterprise at the TechCon time, but Harumi Kuno manages to track me down at least every other year and politely asks me to review again :-). Earlier Career Technologists are indeed one of the nicest contributions of TechCon trailing only the hallway discussions and posters. I too had two ECTs I was responsible for last and previous years and we also had their representation from our team. TechCon, delivered artfully by Carole Macpherson and her team and by many others, such as Julie Christensen in the past, is something all HP(E) people remember. Looking forward to the 25th in a couple of years! Congratulations on your 10 patents Shubham Saloni!Partha Ranganathan liked thisRecently Shubham Saloni Posted about going double-digits with her 10th patent. That and recent posts about HPE TechCon (hello Dejan Milojicic) made me think about the things I had seen there through the years. I had attended every single TechCon from the first one in Keystone Colorado until the year I retired. One year it was decided that Fellows would host a very few ECTs (Early Career Technologists) selected to attend by their business units on the basis of merit. Shubham Saloni was assigned to me. We met over lunch, and it was readily apparent that she was determined to learn as much as possible during the conference. We started with a few formal presentations on technological developments in silicon photonic devices and photonic system architectures that had little to do with her business unit, then went out to the poster sessions. Every formal presentation also had a poster with authors ready to speak about their work. I asked her to speak with the authors throughout duration of the conference, to really dig in and understand what people were doing, why they were doing it, and how it applied to the business. I also asked her for her top three each day. Shubham was on a mission. Every author, every paper. Early mornings till midnight. She spoke with PhD’s about the technology, then the interaction between the theoretical and the practical considerations. When the formal and poster sessions were closed, she asked about how they collaborated on ideas and patents over dinner. By the end of the conference she had notes on every technology, good references for meaningful papers, and had built a solid network of senior technologists that crossed organizational boundaries. I had never seen someone so determined to pursue an opportunity. For the next few years she was my go-to story about why we should include Early Career Technologists in the mix.
-
Partha Ranganathan liked thisI am deeply honored to receive Rising Woman in Semi Award, this week, at #GSA #WISH2025. I feel a mix of gratitude, reflection, and emotion receiving all your wishes and reach-outs. Thank you to GSA and Jodi Shelton to building a space where our voices are not only heard but celebrated. This recognition reflects my long journey at Marvell where I received meaningful opportunities to rise, to lead, and make an impact to the arc of technology and business. Grateful to #marvell, Matt Murphy, Janice Hall where commitment to inclusion and empowerment is not just a policy—it’s a promise by Matt that drives positive changes to semiconductor industry.Partha Ranganathan liked thisCelebrating recognition of leadership and impact in semiconductors. Marvell congratulates Manisha Gambhir, senior director and technology architect, on receiving the GSA Rising Women in Semi Award at the WISH Conference, hosted by the Global Semiconductor Alliance and the GSA Women's Leadership Initiative. The award honors rising leaders who have made exceptional contributions to the growth, innovation and success of the semiconductor industry. With more than 18 years of experience, Manisha leads global teams delivering mass-production-ready mixed-signal IPs for products in data center and AI connectivity, storage and automotive. She also contributes as a thought leader on Marvell innovation, patent and university relations committees, and actively supports mentorship, talent development and women’s leadership initiatives. Her recognition reflects both technical excellence and a commitment to advancing the next generation of industry leaders. Learn more about the GSA Rising Women in Semi Award: https://mrvl.co/48N16ZT
-
Partha Ranganathan liked thisPartha Ranganathan liked thisAfter hosting 400 dinners & 4,000 guests (Olympians, astronauts, CEOs, etc.) The most important truth I learned was: What we are taught about leadership and teams is BS. So I spent three years researching and writing Team Intelligence: How Brilliant Leaders Unlock Collective Genius. Every university, MBA program, and consultancy has told us that there are essential leadership skills. But have you ever noticed that none of the most effective leaders in our society have these skills? • Is Elon Musk known for creating psychological safety or leading through charisma? • Was Steve Jobs known for being supportive and giving good feedback? The only universal characteristic for all leaders is having followers. What gets us to follow them is that they make us feel there will be a better future. It's why when you were in high school, stuck in class on Friday at 1 pm, you felt excited (Your future was freedom). If you can make people feel this, they will follow you. So how do you make people feel that there will be a better future: Lean into your super skills: ↳ Being well-rounded doesn't stand out, but you have a handful of skills that are so disproportionately strong, people will follow them. Help them see their future: ↳ Remember, you are not the hero of their story; they are. Instead, you are the magic tool they can use to create their new future. Avoid breaching the social contract: ↳ The negatives of an employee feeling shame are way worse than the positives of giving them good feedback. Don't try to do it all yourself. ↳ Your job is to lean into what you are great at and let the rest of the team complement your skills. Remember: Leaders don't fit a perfect mold; instead, it’s what makes you special that will cause people to follow you. Embrace your super skills. Now the real question is, what’s the difference between the leader who can build a team and the team that succeeds? To understand that, pick up a copy of Team Intelligence: How Brilliant Leaders Unlock Collective Genius. AMAZON LINK: https://lnkd.in/ebwnt7UP
-
Partha Ranganathan liked thisPartha Ranganathan liked thisOne more thing today: I'm excited to announce a partnership between Data Robot and Aryn! Aryn DocParse now powers the unstructured data prep for Data Robot's Agent Workforce Platform. Together, we deliver the most accurate and reliable agentic workflows for enterprises. See 👇 for more.
-
Partha Ranganathan liked thisPartha Ranganathan liked thisVery proud that Rivos Inc., a Walden Catalyst Ventures portfolio company we helped build from Day one has been acquired by Meta. Congratulations to Puneet, Mark, Belli and Tse-Yu, and the entire Rivos team. It has been a privilege to support you, together with Amarjit, Scott/Dell Capital, Romit/Matrix Capital and my partners at Walden Catalyst on this remarkable journey. In 2021, together with my longtime friend Amarjit Gill, we worked to bring together an outstanding founding team: Puneet Kumar, Mark Hayter, Belli Kuttanna, and Tse-Yu Yeh to start the company. Walden Catalyst Ventures was the anchor founding investor and I served as Chairman to support the team. The journey was far from easy. Rivos faced many challenges, including a lawsuit that created uncertainty in the early years, but the co-founders showed resilience and focus. At that critical period, I deeply appreciate the financial support from my dearest friend David Goel, founder and managing partner of Matrix Capital Management. Along the way, I encouraged a pivot to AI workloads and a closer partnership with Meta, which changed the company trajectory. This outcome is also a reflection of the Walden Catalyst team. I am fortunate to work alongside my partner and co-founder Young Sohn, and our partners Shankar Chandran, Francis Ho, Nicolas Autret, Andy Kau and Roni Hefetz. After building successful companies like Annapurna Labs (acquired by Amazon), Habana Labs (acquired by Intel Corporation) and NUVIA Inc (acquired by Qualcomm), I now want to turn my attention to putting together world class team and building the next generation compute platform from inside Intel Corporation! Full story here: https://lnkd.in/e2uStb-7
-
Partha Ranganathan liked thisPartha Ranganathan liked thisI'm delighted to share that Vercel Closed Series F at $9.3B Valuation to Scale the AI Cloud. Super excited to work with GIC and Khosla Ventures as we grow the AI Cloud. https://lnkd.in/gnAdJWpBVercel Closes Series F at $9.3B Valuation to Scale the AI CloudVercel Closes Series F at $9.3B Valuation to Scale the AI Cloud
-
Partha Ranganathan liked thisPartha Ranganathan liked thisExciting news from our team today with the launch of Tunix! 🎉 Tunix is a high-performance JAX-native library designed for LLM post-training, featuring algorithms for: - SFT (including LoRA, QLoRA) - Preference Tuning (DPO) - Distillation - Reinforcement Learning (PPO, GRPO, GSPO) - And more! For more details, check out https://lnkd.in/gaTGpZpK
-
Partha Ranganathan liked thisPartha Ranganathan liked thisWhen I was in middle school, a college-educated (a big deal in our family then) family friend visited and casually asked me if my school was closed for summer. In India, schools close for summer quite early. I answered, “Yes, it was on March 22,” instead of saying “March 22nd.” Apparently, if you omit the year or read it aloud, you should use “March 22nd.” My slip was met not with compassion but with ridicule. He remarked that to go far in life, one must learn to speak and write English better. It became a personal challenge, one I would carry for years, shaping how I approached learning, communication, and my ambitions. If you have ever wondered how I grew into writing, my journey, narrated below, is the backdrop. I studied in a Tamil-medium government school, where all subjects were taught in Tamil, and English was just one subject among many. My early exposure to English was limited to textbooks and exam preparation. When I entered college, the sudden switch to an entirely English-medium environment felt like stepping into deep water without knowing how to swim. Every lecture was a struggle to understand, and every assignment was a slow battle with vocabulary and grammar. Throughout my college years, I read The Hindu newspaper daily (especially enjoyed Art Buchwald’s satire column), worked through the Wren and Martin grammar book, pored over the English dictionary as if it were a novel, listened obsessively to Voice of America broadcasts on the radio, and devoured The Elements of Style by Strunk and White. Bit by bit, the words stopped feeling foreign, and grammar rules became less like fences and more like scaffolding. The rhythm and nuance of English began to settle in my mind. Years later, I earned a PhD and even became a college professor, instructing students in English in this country, a reality that would have seemed impossible back in those school days. To every student whose mother tongue flows differently from English, your path forward is not only possible but potentially more enriching than for those who began with linguistic privilege. Your multilingual brain already carries remarkable cognitive flexibility, cultural depth, and problem-solving skills that monolingual speakers often lack. The very struggles you face today in translating thoughts, searching for precise words, and navigating cultural contexts are building neural pathways that will serve you throughout your career. Remember that accents carry the music of your heritage, not inadequacy, and that your unique perspective, shaped by a different linguistic lens, offers fresh insights. Your mother tongue is not a limitation to overcome, but a foundation to build upon as you add English to your growing suite of communication tools.
Experience & Education
-
Google
View Partha’s full experience
See their title, tenure and more.
or
Volunteer Experience
-
-
Board Member
LYNBROOK EXCELLENCE IN EDUCATION FOUNDATION
- 6 years 11 months
Education
Secretary/President
Other similar profiles
-
JJ Zhuang
JJ Zhuang
I grew up in Shanghai, came to the US after college, landed in tech by chance, and learned everything by doing. I'm extremely fortunate to have a curious mind, always seeking to understand more, about the physical world, people, and myself. I'm good at solving puzzles, and see every challenge in life as another puzzle to solve. I love working with talented people with passion. It's the best perk of my job.
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Partha Ranganathan
1 other named Partha Ranganathan is on LinkedIn
See others named Partha Ranganathan