Activity
-
Just another day of deceit by Bajaj Allianz General Insurance: After over 10 years of waiting to make the right purchase, we made a decision to buy…
Just another day of deceit by Bajaj Allianz General Insurance: After over 10 years of waiting to make the right purchase, we made a decision to buy…
Liked by Prateek Pandey
-
Excited to unveil #SwiftBharat: AI Amplified, Impact Magnified. This is a unified approach to co-create positive change, bringing together…
Excited to unveil #SwiftBharat: AI Amplified, Impact Magnified. This is a unified approach to co-create positive change, bringing together…
Liked by Prateek Pandey
Experience
Education
Publications
-
Improving Realism of 3D Texture using Component Based Modeling
IEEE Workshop on Applications of Computer Vision (WACV)
3D textures are often described by parametric functions for each pixel, that models the variation in its appearance with respect to varying lighting direction. However, parametric models such as Polynomial Texture Maps (PTMs) tend to smoothen the changes in appearance. We propose a technique to effectively model natural material surfaces and their interactions with changing light conditions. We show that the direct and global components of the image have different nature, and when modeled…
3D textures are often described by parametric functions for each pixel, that models the variation in its appearance with respect to varying lighting direction. However, parametric models such as Polynomial Texture Maps (PTMs) tend to smoothen the changes in appearance. We propose a technique to effectively model natural material surfaces and their interactions with changing light conditions. We show that the direct and global components of the image have different nature, and when modeled separately, leads
to a more accurate and compact model of the 3D surface texture. For a given lighting position, both components are computed separately and combined to render a new image. This method models sharp shadows and specularities, while preserving the structural relief and surface color. Thus rendered image have enhanced photorealism as compared to images rendered by existing single pixel models such as PTMs.Other authors -
Other similar profiles
Explore more posts
-
Pranay Prateek
It was great attending Kubecon NA last month and speaking with Henrik Rexed on his podcast Here's a snippet from our discussion where I talk about what we are building at SigNoz and our Messaging Queue feature which we launched in Kubecon If you use Kafka at scale, do check this feature out. We have leveraged opentelemetry semantic conventions to provide much deeper insights into Kafka and correlating with consumers and producers
2 Comments -
Venkatesh Palani
Presenting at #MRCVegas2024 was a true pleasure and the response to our session, "Future of Strong Auth – Passkey" was fantastic. Shout out to Anand Bahety, Mahendar Madhavan, and the team for their invaluable contribution in enabling passkey capabilities for eBay users. Also, thanks to Kathryn Irwin and Julie Fergerson, CPFPP from #MRC for the opportunity and for making this conference possible. A few things to share on the topic of Passkey and its impact on payment risk management: 1. Passkey has significant potential as an option for 3DS, a standard currently in draft at EMVCo. If anyone is interested in learning or collaborating further, feel free to reach out to Ulrich Herberg, Ph.D. from eBay who has been actively involved in these discussions. 2. It would be particularly intriguing to see how EU regulations will classify passkey—whether as a two-factor or as a single-factor authentication method. 3. I learned from some of my friends about how some EU banks are already adopting mobile apps for verification over SMS. Similar innovations in the U.S. would be awesome! I'm eager to hear your thoughts on the following topics: 1. Has the widespread adoption of 3DS in the EU indeed resulted in lower chargebacks compared to the U.S.? 2. Given Passkey's potential, do you think it has the capability to not only safeguard accounts but also mitigate first-party chargebacks?
-
Mahesh Mallikarjunaiah ↗️
What happens under the hood when you turn on your Android mobile phone When you power on your mobile phone, a meticulous self-test is conducted on all the hardware components. This comprehensive check ensures that every part of your device is functioning flawlessly. Once this rigorous hardware test is successfully completed, the Boot ROM, stored on the CPU, initiates the loading process. The bootloader loads Android OS, validating certain aspects of the initial loading. The kernel loads device drivers and background processes and manages the system resources. The init process loads and mounts directories and files and starts various processes. The Android runtime is initialized and is mainly used by the Android operating system. As the runtime, ART executes the Dalvik executable (DEX) format and DEX bytecode specification. After these system services are started, the activity manager kicks in to bring the user the initial screen to start interacting. Power-On Self-Test (POST): When you press the power button, the phone undergoes a quick self-check process known as POST. This test ensures that essential hardware components like the processor, memory, and battery function correctly. Bootloader: After the POST is successfully completed, the bootloader, a small yet crucial program stored in the phone's read-only memory (ROM), takes charge. It not only initializes the hardware but also orchestrates the booting process, including selecting the operating system to load. Loading the Kernel: The bootloader loads the Android kernel into the device's RAM (Random Access Memory). The kernel is the core of the Android operating system. It manages the device's resources, such as the CPU, memory, and input/output operations. Init Process: Once the kernel is loaded, the init process (short for initialization) is started. This process is responsible for initializing the Android system and launching essential system services and daemons. Android Runtime (ART) Initialization. This is a critical step as ART is the runtime environment where Android applications come to life. It's responsible for a significant task and compiles the application code into machine code, ensuring efficient execution on the device's hardware. This makes ART a crucial part of the booting process. Starting System Services: After the initialization of ART, various system services are started. These services include the Android System Server, which manages core system components like the package manager, activity manager, and window manager. Launcher and User Interface: Once the system services are up and running, the launcher (the user interface that displays your home screen, app icons, and widgets) is started. This allows you to interact with your device and access your installed applications. Loading Applications: Any pre-installed applications and services configured to start automatically are loaded. These may include system apps like the dialer, messaging app.
21 Comments -
Mahesh Mallikarjunaiah ↗️
What happens under the hood when you turn on your Android mobile phone When you power on your mobile phone, a meticulous self-test is conducted on all the hardware components. This comprehensive check ensures that every part of your device is functioning flawlessly. Once this rigorous hardware test is successfully completed, the Boot ROM, stored on the CPU, initiates the loading process. The bootloader loads Android OS, validating certain aspects of the initial loading. The kernel loads device drivers and background processes and manages the system resources. The init process loads and mounts directories and files and starts various processes. The Android runtime is initialized and is mainly used by the Android operating system. As the runtime, ART executes the Dalvik executable (DEX) format and DEX bytecode specification. After these system services are started, the activity manager kicks in to bring the user the initial screen to start interacting. Power-On Self-Test (POST): When you press the power button, the phone undergoes a quick self-check process known as POST. This test ensures that essential hardware components like the processor, memory, and battery function correctly. Bootloader: After the POST is successfully completed, the bootloader, a small yet crucial program stored in the phone's read-only memory (ROM), takes charge. It not only initializes the hardware but also orchestrates the booting process, including selecting the operating system to load. Loading the Kernel: The bootloader loads the Android kernel into the device's RAM (Random Access Memory). The kernel is the core of the Android operating system. It manages the device's resources, such as the CPU, memory, and input/output operations. Init Process: Once the kernel is loaded, the init process (short for initialization) is started. This process is responsible for initializing the Android system and launching essential system services and daemons. Android Runtime (ART) Initialization. This is a critical step as ART is the runtime environment where Android applications come to life. It's responsible for a significant task and compiles the application code into machine code, ensuring efficient execution on the device's hardware. This makes ART a crucial part of the booting process. Starting System Services: After the initialization of ART, various system services are started. These services include the Android System Server, which manages core system components like the package manager, activity manager, and window manager. Launcher and User Interface: Once the system services are up and running, the launcher (the user interface that displays your home screen, app icons, and widgets) is started. This allows you to interact with your device and access your installed applications. Loading Applications: Any pre-installed applications and services configured to start automatically are loaded. These may include system apps like the dialer, messaging app.
27 Comments -
Ahsan Nabi Dar
The more I work with #Elixir the more I come to appreciate the language and its #community. We at DAT AI Systems Pvt Ltd did a major migration from #Kafka to #AMQP without any downtime or client impact 💪. #Queues are the unsung heroes of #reliable, #scalable, and #distributed systems. With the AMQP protocol, an open standard widely adopted across industries, you can easily integrate #messaging into your stack without worrying about vendor lock-in. Why should you use queues in your architecture? - Decouple services for better scalability and resilience - Handle #asynchronous workloads efficiently - Ensure message #reliability and guaranteed delivery - Benefit from strong AMQP support across cloud and on-prem environments As businesses grow, embracing #open standards like AMQP ensures #flexibility, #portability, and long-term #stability. Ready to #future-proof your system? Check out my latest blog on the benefits of queues and AMQP. dev.to: https://lnkd.in/gDvce_nb Medium: https://lnkd.in/gVp9euNu #AMQP #MessageQueues #CloudArchitecture #Scalability #OpenStandards #EnterpriseTech #DistributedSystems
3 Comments -
K S AKSHAY
A Day at the Apache Kafka 🚀 Today, I had the pleasure of attending an in-person Apache Kafka hosted by CRED in Bangalore. Here’s my feedback on this incredible event: 📍 Venue: Cred Aero, Bengaluru 🔹 Agenda Highlights: - @Vikas Sharma , CRED : Talk: Running Kafka Optimally in Production: Practical Usage Feedback: Vikas shared practical insights on optimizing Kafka in production environments, with valuable strategies for resource management. - Udayaram Kammara , Bazaarvoice : Talk: How Bazaarvoice Utilizes Kafka for Both Stream and Batch Processing Scenarios Feedback: Udayaram's session on Kafka's use at Bazaarvoice was enlightening, covering real-world use cases and challenges. - Diptiman Raichaudhuri , Confluent : Talk:Stream Processing for the Non-JVM World: pyflink and kafka-python Feedback: Diptiman's explanation of Apache Flink's architecture and the demonstration of kafka-python and pyflink were fascinating and beneficial for non-JVM users. What I Learned: - Effective strategies for running Kafka in production. - Real-world use cases and challenges in stream and batch processing. - Leveraging pyflink and kafka-python for stream processing. Overall, this meetup was a treasure trove of knowledge and inspiration. The sessions were well-organized, and the speakers were exceptionally knowledgeable. I’m excited to implement these learnings in my projects and continue exploring the potential of Apache Kafka and related technologies. 🙌 #ApacheKafka #KafkaMeetup #CRED #Bangalore #EventStreaming #StreamProcessing #TechCommunity #Networking #Python #Flink #DataPlatforms
2 Comments -
Nishant S.
My first ever article is now published on the Google Cloud Community Blogs, and I couldn’t be happier! 💫 We worked on splitting Looker instances, a case for multiple production instances, which was both innovative and challenging. This approach significantly improved the performance, which inspired us to document it as an article and share it with the broader Looker community. 🚀 A huge shoutout to my ex-team at Cimpress and special thanks to Maryia Borukhava for her invaluable collaboration and support in writing this article. 🙂 Check it out here: https://lnkd.in/dVdmt3Yf #Looker #Google #Article
16 Comments -
Sandeep Deshmukh, PhD
As TeemGenie turns 4, we decided it was time to get a little candid. Here’s the first of many snippets from my conversation with Devayani, where we talk about some key insights from our journey. In this clip, I dive into the unique challenges of hiring for VC-funded startups building enterprise products, and how it’s different from traditional hiring. Finding the right people in this space is key to helping these startups scale effectively. Looking forward to sharing more in the coming episodes! #TeemGenieTurns4 #FlyHighTeemGenie #CandidConversations #StartupHiring #ScalingStartups
2 Comments -
Sumon Mal
🚀 Transforming OTT Platforms with MongoDB Atlas: A Game-Changer for SonyLIV At MongoDB.local Bengaluru 2024, I had the privilege of sharing how SonyLIV revolutionised its CMS platform, Blitz, leveraging MongoDB Atlas and its powerful search capabilities. The results? A 98% reduction in search query latency and seamless scalability for 495,000+ documents! Here’s what made it possible: ✅ MongoDB’s flexible document model for handling dynamic, large-scale video files. ✅ MongoDB Atlas’ Online Archive feature for optimised performance and precise datasets. ✅ A collaborative effort between SonyLIV’s engineering team and MongoDB’s expertise to unlock unparalleled value. This is a testament to how the right technology, combined with expert guidance, can accelerate innovation and deliver flawless customer experiences. 📖 Check out the full blog and session recording to dive deeper into this transformative journey. #MongoDB #OTTInnovation #TechTransformation #SonyLIV #MongoDBAtlas #CloudDatabase #EngineeringExcellence
-
Aryaman Maheshwari
Is the hype around going into an Old IIT still relevant? or was there no difference to begin with? are there benefits to new IITs which old one's just cannot provide for? More discussions and fun light hearted conversation in this part of the podcast!. Hope you guys enjoy it! https://lnkd.in/gw-CmFra
-
Sirish Puppala
Nothing excites me more than seeing the passion, grit, and #innovation that startup founders bring to the table. Thank you Atal Incubation Centre, AIC-GNITS Foundation, GNITS_IT, Silverneedle Ventures & Dr Shashwat Pathak for the opportunity and pleasure sharing the jury panel with seasoned investor founder Ravi Teja Guptaji @founders: Keep #iterating, #learning, and #growing. This is just the beginning of your journey! 🚀
-
Abhigyan Suman
At NVIDIA’s AI Summit, I met Anshul Khandelwal, Co-founder and CTO of invideo, and attending his insightful session on their lip-sync model. They serve 25 million users, the model leverages NVIDIA’s StyleGAN with audio-driven style learning to deliver realistic, high-quality lip sync while maintaining facial identity, which stood out. Anshul also shared the challenges of scaling the model and the key lessons learned during its development. To efficiently handle their large user base, they utilize TensorRT and Triton Inference Server to optimize performance and reduce costs. The session dived into how AI is transforming video content creation, empowering creators to produce high-quality multimedia experiences with ease. It was inspiring to learn how InVideo’s platform, powered by E2E Cloud Network’s GPUs, is driving innovation in digital content. Thank you, Anshul, for sharing your journey and insights! It was a pleasure meeting you. Hoping to stay connected. Cheers! #invideo #nvidia #ai #virtualhumans #dcverse
-
Supratik Das
Recently attended droidcon with Yash Khandelwal and would've definitely regretted not witnessing it in person. → Gained some concrete validation from small and big teams alike on how deep the Android's dev productivity challenges are → It's always exciting to see how big the #Android ecosystem has come to be, with not just phones but also #wearables, #smartglasses, #securitysystems, and #AI-enabled multimodal handhelds, and how those same productivity and collaborative challenges propagate largely here → A smart platform with relevant tooling is a great enabler for faster delivery, which we here at Dashwave are striving to bring forth PS: I am in SF for the next few weeks, HMU to say hi, catch up or chat on, but not limited to, dev tools, Android engineering, distributed systems and webassembly.
-
Pramodh BN
"𝐁𝐫𝐚𝐜𝐞 𝐟𝐨𝐫 𝐢𝐦𝐩𝐚𝐜𝐭" 𝐰𝐡𝐞𝐧 𝐃𝐡𝐨𝐧𝐢 𝐚𝐫𝐫𝐢𝐯𝐞𝐬: 𝐇𝐨𝐰 𝐉𝐢𝐨𝐂𝐢𝐧𝐞𝐦𝐚 𝐄𝐧𝐬𝐮𝐫𝐞𝐬 𝐅𝐥𝐚𝐰𝐥𝐞𝐬𝐬 𝐈𝐏𝐋 𝐒𝐭𝐫𝐞𝐚𝐦𝐢𝐧𝐠 𝐭𝐨 𝟐𝟎 𝐌𝐢𝐥𝐥𝐢𝐨𝐧 𝐅𝐚𝐧𝐬 "𝐁𝐫𝐚𝐜𝐞 𝐟𝐨𝐫 𝐢𝐦𝐩𝐚𝐜𝐭," is a term that JioCinema tech team uses when Dhoni enters. - When Dhoni arrives, spikes are gradual - Playback system and playback API's get a spike - When Dhoni gets out, all the APIs on the home page gets a spike as all are exiting the live stream and going to watch something else When millions of IPL fans tune in to watch their favorite teams, the stakes are incredibly high. At JioCinema, the tech team understands the passion and excitement that cricket brings. Here’s a glimpse into the meticulous preparation and innovative solutions that make it possible, shared by 𝐏𝐫𝐚𝐜𝐡𝐢 𝐒𝐡𝐚𝐫𝐦𝐚, 𝐒𝐫 𝐄𝐧𝐠. 𝐃𝐢𝐫. 𝐚𝐭 𝐉𝐢𝐨𝐂𝐢𝐧𝐞𝐦𝐚. 𝐔𝐧𝐝𝐞𝐫𝐬𝐭𝐚𝐧𝐝𝐢𝐧𝐠 𝐅𝐚𝐧𝐬' 𝐄𝐦𝐨𝐭𝐢𝐨𝐧𝐬: The IPL is more than just a sporting event, it's an emotion that unites millions. Each match is a high-stakes event. 𝐒𝐭𝐢𝐜𝐤𝐞𝐫 𝐒𝐡𝐨𝐜𝐤: 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞: A new sticker-sharing feature caused a surge of 1 million requests per second on CDN. 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧: The team optimized cache TTL and bundled stickers into a single API call, reducing requests from 10 million RPS to 200K RPS. 𝐃𝐍𝐒 𝐃𝐢𝐥𝐞𝐦𝐦𝐚: 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞: Users on a specific ISP couldn't access the app due to DNS resolution failures. 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧: Google's public DNS and a fallback DNS resolver were implemented within the app. 𝐒𝐜𝐚𝐥𝐢𝐧𝐠 𝐃𝐚𝐭𝐚𝐛𝐚𝐬𝐞𝐬: 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞: Handling unexpected traffic spikes, especially during key moments in the match. 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧: The databases were pre-scaled based on detailed calculations and increased cache TTLs. They also implemented panic modes with static responses to ensure continuity even under heavy load. 𝐂𝐃𝐍 𝐌𝐚𝐧𝐚𝐠𝐞𝐦𝐞𝐧𝐭: 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞: Managing multiple CDNs to handle massive traffic and ensure redundancy. 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧: The multi-CDN optimizer dynamically switches traffic based on real-time performance, ensuring seamless streaming throughout the match. 𝐀𝐝 𝐈𝐧𝐬𝐞𝐫𝐭𝐢𝐨𝐧: 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞: Delivering targeted ads without disrupting the viewing experience. 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧: Use server-side ad insertion with cohort targeting to ensure personalized ads reach the right viewers without affecting stream quality. 𝐏𝐫𝐞𝐩𝐚𝐫𝐚𝐭𝐢𝐨𝐧 𝐚𝐧𝐝 𝐏𝐥𝐚𝐧𝐧𝐢𝐧𝐠: Preparation for the IPL starts in Nov if the season is from march, with extensive tests and audits to ensure system resilience. From database scaling to CDN optimization, every aspect is fine-tuned to handle the unique challenges each IPL season brings. Watch the full video here: https://lnkd.in/gsivAuzN #IPL2024 #LiveStreaming #TechInnovation #EngineeringExcellence #JioCinema #TechChallenges #CricketPassion #SeamlessStreaming #BackendEngineering #CDNOptimization #DatabaseScaling #AdTech
-
Shadi Copty
Evaluating my fine tuned 3B model with LLM judge - many of you have asked so here goes :-). Tl;Dr - 91% accuracy vs 47% not-fine tuned; with only 140 examples and 3 minutes of training (https://lnkd.in/gHw993ZS) I started out with creating a new synthetic test data set that the fine tuned models haven't seen, using the same code from the training set (data here: https://lnkd.in/gcCSdT5e). I used Comet's Opik per Armand's recommendation (you should follow him if you aren't :-)). Then built my evaluations test for (1) JSON schema compliance, (2) distance from reference answer (3) the LLM Judge which scores first on json format, then on compliance with schema, then on entities being detected, then on not detecting more than necessary. I tried this out with a bunch of LLMs that could run on this laptop, as you can see the 3B fine tuned one performed the best, followed by the 1B, then all of the out of the box ones pretty similar and not so great. Code for you to enjoy: https://lnkd.in/guhzdUkf You'll need a Comet account, which is shockingly free. Enjoy :-) Next I'm going to try to beef up the data gen with Distilable and CrewAI, let me know if you find that interesting or have other ideas.
19 Comments
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore MoreOthers named Prateek Pandey in India
-
Prateek Pandey
Business HR | HR Operations
-
Prateek Pandey
-
Prateek Pandey
-
Prateek Pandey
Data Engineer at Capgemini | Snowflake | Azure
-
Prateek Pandey
810 others named Prateek Pandey in India are on LinkedIn
See others named Prateek Pandey