🪄 Blending Realities | 🎙️ Host, TED AI Show | 🚀 Scout, A16z | 🎬 Creator w/ 1.4M+ Subs & 450M+ Views | 🌎 Ex-Google PM, 3D Maps & AR/VR 🥽 Bilawal Sidhu is a creator, engineer, and product builder obsessed with blending reality and imagination through art, science and storytelling. With more than a decade of experience in the tech industry, he spent six years as a product manager at Google, where he worked on spatial computing and 3D maps. His work has been featured in major publications including Bloomberg, Forbes, BBC, CNBC, and Fortune, among others. Bilawal’s journey into computer graphics began at 11, when he fell in love with seamlessly blending 3D into real life footage. Since then, he's captivated over 1.4M subscribers, garnering more than 450M views across his platforms. Driven by a mission to empower the next generation of artists and entrepreneurs, Bilawal openly shares AI-assisted workflows and industry insights on social media, demystifying AI creation and inspiring others to embrace the potential of the "co-pilot era." He is also an angel investor and a16z scout, focusing on generative AI and spatial intelligence startups. An alumnus of the University of Southern California, Bilawal holds a dual degree in Computer Science and Business Administration. Growing up across India, the UK, and the US, he's fluent in Hindi and Punjabi. When not exploring the frontiers of technology, he practices mantra meditation, blending ancient wisdom with modern insights, and collects Ibanez and Ernie Ball electric guitars. Please email: bilawal@metaversity.us for business opportunities and speaking engagements. Positions: Scout at Andreessen Horowitz (2024 - Present), Tech Curator & Host of The TED AI Show at TED Conferences (2023 - Present), Creative Technologist & Creator (1.4M+ Total Connections) at Metaversity LLC (2023 - Present), Senior Product Manager, Immersive View (AI, 3D, NeRF, Cloud Streaming) at Google (2019 - 2023), Product Manager & Producer (360/180 VR, Light Field & 3D Media, AR Platform SDK) at Google (2017 - 2019), Consultant, Digital Business & Strategy Group at Deloitte Digital (2014 - 2016), Analyst, Emerging Technologies Group at Deloitte Digital (2014 - 2014), Business Technology Analyst at Deloitte Consulting (2011 - 2012), Global Strategy Intern (Business Performance Services) at KPMG Advisory (2010 - 2010) Skills: Artificial Intelligence (AI), Product Management , Computer Graphics, Computer Vision, Generative AI, Visual Effects, Podcasting, Science Communication, Technology Journalism, Digital Strategy, Virtual Reality, Management Consulting, Product Design, Autodesk Maya, Adobe After Effects, Information Architecture, Rapid Prototyping, Business Intelligence, 3D Printing, Data Warehousing, Remote Sensing, Research and Development (R&D), User Research, Photoshop, Adobe Premiere, HTML 5, CSS, Adobe Creative Suite, Unity3D, Unreal Engine 4, YouTube, Leadership, Public Speaking, Interactive Media, 3D Studio Max, Augmented Reality, Visual Design, Digital Compositing, Oculus Rift, Virtual Production, Mixed Reality, Stereo 3D, Photogrammetry, Light Fields , Volumetric Content, Augmented Reality (AR), 3D Reconstruction, SLAM, Short form video, Tiktok Recent Posts: The lines between code & content are blurring. I made this 3d city block animation in claude 3.7 sonnet. Then I used runway's video-to-video AI to style it like a lego city at night. At the rate things are going, this'll be a shader running in real-time. Blender-Made Movie Flow Wins an Oscar at the 97th Academy Awards Ceremony. This is a milestone in computer graphics history — the open source marvel that is Blender will only be more ascendant from here. Perhaps now, we can finally say Blender is “industry-standard”? #3d #cgi #vfx Wild to see 3d rendering & interactive experiences slowly being eaten by AI video. Research from big labs is even convincing skeptics. Here are 3 must-read papers: 1. Microsoft Muse: a "World and Human Action Model" that can "generate game visuals and controller actions." 2. Google Deepmind Genie 2: a "foundation world model" capable of "generating action-controllable, playable 3D environments." 3. Hedra/Nvidia's Magic 1-For-1: real-time generative video model that can "generating one minute video clips within one minute It is no surprise that startups like Luma and Odyssey have shifted from 3D-first to generative video to realize their world-building dreams. Bookmark this post and tell me if these research projects convince you that we are quickly heading towards a future where "every pixel is generated, not rendered." Wow. Recreating the Shawshank Redemption prison in 3D from a single video, in real time (!). Just read the MASt3R-SLAM paper and it's pretty neat. Here's the TL;DR: → These folks basically built a real-time dense SLAM system on top of MASt3R, which is a transformer-based neural network that can do 3d reconstruction and localization from uncalibrated image pairs. → The cool part is they don't need a fixed camera model -- it just works with arbitrary cameras -- think different focal lengths, sensor sizes, even handling zooming in video (FMV drone video anyone?!). If you've done photogrammetry or played with NeRFs you know that is a HUGE deal. → They've solved some tricky problems like efficient point matching and tracking, plus they've figured out how to fuse point clouds and handle loop closures in real-time. → Their system runs at about 15 FPS on a 4090 and produces both camera poses and dense geometry. When they know the camera calibration, they get SOTA results across several benchmarks, but even without calibration, they still perform well. → What's interesting is the approach -- most recent SLAM work has built on DROID-SLAM's architecture, but these folks went a different direction by leveraging a strong 3D reconstruction prior. Seems to give them more coherent geometry, which makes sense since that's what MASt3R was designed for. → For anyone who cares about monocular SLAM and 3D reconstruction, this feels like a significant step toward plug-and-play dense SLAM without calibration headaches -- perfect for drones, robots, AR/VR -- the works! Link to code release and paper in comments below. With just a few prompts I made this multi-camera city block simulation in threejs. Not sure what i'm doing, but i think they call it "vibe coding" with Claude 3.7 Sonnet. It genuinely feels like the lines between code & content are blurring. One of many fun things i'm stoked to cover in this week's video; link to channel below. #ai #3d #cgi #geoint #anthropic Turn your phone into an instant 3D floor plan creator with Apple's RoomPlan API in Polycam. This computer vision technology doesn't just map your space — it understands it, showcasing the groundwork Apple laid for Vision Pro and spatial computing devices to follow.​​​​​​​​​​​​​​​​ #3d #ai #computervision Anthropic after dropping Claude 3.7 Sonnet. Turn sound on. So I think Apple is actually playing this very smart. But this isn’t news or surprising. This is the first year that my Pixel 9 Pro feels *better* than my iPhone 16 Pro. Never thought this day would come, but here we are, especially given how good the Gemini AI integration is on Android vs. the absolute dumpster fire that is Apple Intelligence (I literally turned it off). Plus, a win for Google users like me who use iOS but are on Gmail, Docs, Calendar, Maps etc. But let’s not forget that Apple always said they were going to bring Gemini in the future — Craig Federighi even said it back at WWDC last year, saying they’d do “future integrations,” explicitly calling out Google Gemini. Apple’s gonna focus on “semantic index,” creating embeddings of all your data and allowing you to load context on the fly, and use whatever AI model you need to get the job done — whether it’s one of their in-house variants (incidentally trained on Google’s TPUs) or a third-party model from OpenAI, Google, or Alibaba (Apple’s model provider of choice for Chinese users). I think a strong partnership model built around their stellar distribution is the only way Apple can catch up, given they’ve focused on AR/VR vs. generative AI for the last few years. Nothing like getting your BBQ aerially bombarded into your backyard — and sometimes your neighbors 😭 This is Zipline, a California-based drone delivery company that has recently started operating in the Dallas-Forth Worth area. Vibe coding is how creative builders actually work. It's the fusion of technical skill and artistic intuition — writing code and wrangling pixels, while surrendering to serendipity. It’s wu wei — effortless action in the digital realm. A movement was inevitable, and it took Karpathy to codify it. #ai #tech #creative New video up! Want to know which AI tools are actually transforming 3D and VFX? In this deep dive, I cover everything from NeRFs and Gaussian Splatting to AI segmentation and object tracking, multimodal LLMs, and next-gen VFX workflows — focusing on tools you can use TODAY. Link: https://lnkd.in/ge8VzFEt #ai #3d #vfx #cgi #computervision The internet going wild with the microwave AI filter -- probably because it's pure nightmare fuel 😭 If you want to give this a try look up the template on CapCut -- pretty funny application of computer vision tech -- wherein the failure cases are actually a feature, not a bug! #ai #3d #vfx Bro is literally driving around a 3d gaussian splat of a city block. Layer in procedural 3D elements ala. Houdini and an AI video model like NVIDIA Cosmos and you could really have an interesting world simulator with the best of reality capture & game engine technologies. Uses hierarchical 3d gaussian splatting, link in comments to full clip. GPT-4.5 AI's wild week: Musk's bid for OpenAI, Vance at Paris AI Summit, Palmer's $22B VR headset. Video advances with Google's Veo 2, Adobe Firefly Video, and Luma Labs Ray 2. LiDAR uncovers another ancient city, and so much more. Watch the full 40 minute episode in the link below. 🔗 https://lnkd.in/gmseicYr Alibaba strikes again. Full-body swap anyone in a video with just a photo reference. What’s wild to me is that this tech completely bypasses the 3d pipeline (i.e. what Wonder Dynamics does to accomplish similar output) and yet looks so damn good. Basically viggle on steroids. Paper: Animate Anyone 2 (link in comments below) #ai #vfx #cgi #ml Surprise. Google Veo 2 is now available inside the YouTube app! The $0 VFX studio is here. I gave it a test drive; here's what you need to know: > Wayyyy faster than the standalone version of Veo 2 (likely a distilled model optimized for serving?) > Can use Veo 2 clips as "green screen" background or standalone clip > Uses a text-to-image-to-video flow and includes a few preset styles (vintage, anime, clay, digital etc) > Pick your images or re-roll, and then proceed to image to video step > Can add multiple clips in the timeline to make a longer short > Only for US, Canada, Australia, and New Zealand users at the moment Pretty darn exciting to see this in the wild. I think shorts creators will do fun stuff with it -- bringing your smartphone one step closer to a VFX studio in your pocket! #ai #vfx #ml #aivideo #cgi Robotics + AR/VR is a match made in heaven. No better way to share spatial context, as demonstrated in this video by Meta's AI research team. This video is also a glimpse into what Anduril can accomplish now that they’re taking over the IVAS (US military XR headset) project. The future is multi-modal and multi-domain. #ai #robotics #augmentedreality #virtualreality #machinelearning Typical Cursor user: “No, I said add it to the existing drop down menu, not a new one” #ai #coding #llm #programming Hot take: Imagen 3 is now better than MidJourney and FLUX. Give it a try in the Google Gemini app or FAL — it’s a ton of fun. #artificialintelligence #machinelearning #aiart That weird feeling as a creator where every video you upload to a social media platform is effectively training your own AI replacement. This new AI paper is called Goku -- ByteDance research has introduced new family of models that are great at making ads. TL;DR: It's a transformer-based architecture and jointly trained on image + video data. Plus, a finetuned variant for advertising called Goku+ which enables product placement in any scene. "UGC" content creators are cooked. Why? Because tech like this completely disintermediates them -- enabling small businesses to make authentic vertical video advertising without human "UGC creators" in the loop. The writing has already been on the wall. Building your own island of influence before the barrage of fully generative content might be the only moat creators have left. As always, link to paper/project in comments below. #ai #marketing #socialmedia #creators Late to the party, but 2024 was pretty decent! Excited for what this year will bring. New video up! 7 groundbreaking tools and research projects that are transforming AI video creation. Plus, a deep dive into where 3D + video is headed next. Topics covered include: > Pikadditions & Pika 2.1 > OmniHuman-1 (ByteDance) > VideoJAM, SAMURAI & Movie Gen (Meta AI) > MidJourney CEO on Future of 3D & AI Video > Topaz Starlight Upscaling > MoCapade 3.0 Motion Capture Link to full video in comments below. #ai #3d #vfx #cgi I don't know about y'all but this triggered my PTSD. This isn't a big tech parody, it's a dang near documentary lol 1.9M views on this AI generated music video. ViVO instead of VEVO. Uses the likeness of Justin Bieber and Jung Kook without their permission. This is gonna get messy. This is made with a video clip + one reference image in Pika Labs. I shudder to imagine how long these VFX shots would’ve taken the “classical” way… 3d camera tracking, lighting estimation, occlusion handling, asset creation, animation, compositing etc. Now you just need a reference image of the object or character you want to insert, along with a prompt and AI handles the rest. VFX is becoming so accessible, it kinda blows my mind 🤯 #ai #3d #vfx #cgi Heading to SXSW? Catch me on the panel at the 2025 Austin AI Film Festival, March 14-15! I'll be alongside talent from Secret Level, Hollywood, and the AI creator scene. Grab your tickets below — early bird pricing ends Feb 15! Get tickets here: https://lnkd.in/gJzCYKw2 Google AI just dropped major updates to Gemini. Here's the TL;DR: > Gemini 2.0 Pro (woot!) launched as an experimental model in AI Studio, Vertex AI & Gemini Advanced app. > 2.0 Flash Thinking lands in Gemini app with YouTube/Search/Maps integration > Gemini 2.0 Flash now available to developers after last week's consumer release > Flash-Lite introduced as the new budget-friendly option, matching 1.5 Flash pricing (!) continuing the race towards intelligence too cheap to meter. #ai #llm #ml #google “But AI will never get good at motion and physics” VideoJAM: hold my beer… *Introduces motion priors, outperforms Runway, Sora & Kling* Oh and works on any video model with minimal tweaks. *Link to AI research paper in comments below #ai #ml #vfx #cgi I'm honestly kind of in awe of how good AI video is getting lately. This is Google Veo 2, and the output is nearly indistinguishable from reality. Seriously mind blowing at times. Almost unnerving to see how good this looks. What even is real anymore? Audio added using MMAudio. #ai #vfx #cgi Video-to-video AI is underrated. Stop wasting your time text prompting, and re-rolling the slot machine just to get the right composition. 3D is easy -- you can literally greybox & kitbash your scene together in minutes. Can't wait till this tech is running in real-time. #ai #vfx #3d #cgi AI & 3D tools are a blast when you have a VFX channel. Also grateful that my new AI-focused YouTube channel (Bilawal Sidhu) has crossed 10K subs. Here's my 2025 creation stack: > Google Veo 2, OpenAI Sora (text-to-video) > RunwayML (video-to-video) > Kling AI, Minimax, Luma (img-to-video) > Krea AI & FAL (for FLUX & fine tuning) > Topaz Video AI & Magnific (upscaling) > Eleven Labs, Cartesia (voice & dialogue) > Suno, Udio, MMAudio (sound & music) > PostShot & Polycam (3d gaussian splats) > Skyglass VFX (unreal engine virtual sets) > Photoshop, After Effects, Premiere (editing) > Mocha Pro, MaskPrompter, QuickDepth (AI plugins for Adobe tools) > DeepSeek R-1 (reasoning), Gemini 2.0 Flash (long context), Claude (vibes) Personalized, just-in-time AI content is here. NotebookLM was just an appetizer. Audio today, multimodal media tomorrow — it'll be like having a world class newsletter or video podcast made just "for you” Link to enable in comments below Adobe Generative AI Leaders Summit. The talent assembled here is insane; grateful to be a part of it 🙏 Many creators use custom GPTs and Claude projects to generate & refine prompts for their AI image & video creation workflows. But what if you could do it all in one blazing fast user interface? Krea AI has answered that question, and it’s powered by DeepSeek: Is anyone really surprised that OpenAI saying DeepSeek distilled their intellectually property without permission isn’t garnering any sympathy whatsoever? "90% of my LinkedIn content in 2024 was wrong." That's the hard truth I learned today at a16z speedrun One hour with Bilawal Sidhu (Ex-Google PM turned creator with 1M+ followers) and Eric Lay (who's helped hundreds of B2B founders nail their content strategy) completely changed my perspective. My 2024 numbers tell an interesting story: → 3K → 10K followers → 627,835 impressions (+3,461.4% growth!) → 137,592 members reached → Consistent daily engagement See those spikes in July/August? That's when we launched our MVP. Here's what I learned: ↳ Product launches hit different when you've built an engaged audience ↳ Consistent content before launch = bigger launch impact ↳ Your audience becomes your first champions Even with these "successful" metrics, I was still doing it wrong: ↳ 90% of my content = startup life, founder journey ↳ 10% = actual customer problems Today's session changed everything. Here's the gold: 1. The B2B Content Truth - You don't need 100K followers - You need 100 right people engaging - Quality > Quantity - Consistency > Perfection 2. Video Content That Converts - Vertical: Thought leadership, personal brand - Horizontal: Product announcements, demos - Raw screen recordings > polished videos - Tools progression: ⇢ Level 1: Quick captures > Loom (great for starters!) ⇢ Level 2: Pro recordings >Screen Studio (solid features) ⇢ Level 3: Advanced editing: Premiere + TimeBolt 3. LinkedIn Algorithm Secrets - 30-min engagement window (before/after posting) - Comment on ICPs' posts first - Turn on post notifications for decision-makers - Focus on shares > likes - Target second-degree connections 4. Content Creation Framework - Record customer conversations - AI note-taking (game-changer!) - Transform insights into content - Test different ICPs systematically 5. The Psychology of B2B Content - Decision-makers want education, not pitches - Build trust through consistency - Share failures alongside wins - Focus on their problems, not your features 6. Platform Strategy - Twitter → Threads → LinkedIn - Adjust tone per platform - Cross-pollinate without copy-paste - Build once, distribute smartly The biggest mindset shift: Stop treating content as marketing. Start treating it as a conversation. Building Layerpath taught me this: Great demos don't just show features - they tell stories. They create conversations. They build trust. That's why we're helping B2B founders turn their product stories into demos and videos that actually convert. If you're: ⇢ Struggling with demo creation ⇢ Wrestling with product-led content ⇢ Want to make content that converts DM me. Let's chat about how we can help! 2025 Goals: More customer pain points Less generic startup content Focus on specific solutions Build in public, but with purpose What's your biggest B2B content challenge? Drop it below 👇 In SF this week for an a16z speedrun talk and then Adobe’s GenAI leaders summit. Hit me up if you want to kick it. DMs are open. How it’s going: DeepSeek #1 on App Store charts in the USA. Funnily enough #6 is Temu. Include TikTok and we're watching Chinese tech shape American AI, entertainment, and e-commerce in real-time. China rises as DeepSeek ascends to #3 in Productivity and #9 overall on the Apple App Store charts in the U.S.A. Feels like the Sputnik moment for AI. NEW VIDEO: Hands-on review of OpenAI Operator, their first AI agent that can control your browser. As OpenAI moves away from open source, Chinese labs like DeepSeek and ByteDance are becoming the new champions of open AI development, and doing it with access to a fraction of the compute available to US labs. In this deep dive video, I get into > Hands-on demo of OpenAI Operator performing real-world tasks > Claude Computer Use & Google's Project Mariner > DeepSeek R-1 and ByteDance UI-TARS > The evolution from AI agents to AR/VR and robotics > Privacy & security implications of agentic AI Link in comments below #ai #aiagents #agenticai #openaioperator #deepseek #ml #technews > Stargate: our AI data center will be the size of Central Park > Zuck: Oh yeah? Ours will be the size of Manhattan Using OpenAI Operator to make internet memes. Because obviously this is highest utilization of AI agents. Spatial Computing is effectively turning into a three way battle between Meta, Apple and Google. Meta has of course been carrying the VR space, but now Google has partnered with OEMs like Samsung and HTC to bring Android XR to the party. And this move shows Google is serious about VR this time (not just AR). Apple seems to be going after the premium market, leaving Google and Meta to duke it out for the mass market. #augmentedreality #virtualreality #spatialcomputing #ai 1) we’re so back 2) we’re so cooked When watching this demonstration of robotic prowess you can feel humanity’s peak and humanity’s last chapter vibes simultaneously. This is an embodied AI company DeepRobotics based in China. Well that was quick. Could’ve dragged it out for a few days, but I guess the gen alpha lobby has serious pull. When a news outlet claims to have "undeniable proof" of an egg-shaped UAP and crash retrieval programs -- here's what I expect: Instead, we got something deeply uncompelling (link in comments). So, I went wild with OpenAI Sora and little bit of Veo 2 to change that. #ai #vfx #aivideo #cgi Aakash Gupta's latest podcast episode with Bilawal Sidhu is a must-listen for anyone in tech. It’s packed with actionable insights that transcend job roles and career stages. My three favorite insights - 1. Chase the work you enjoy doing, not the titles. When you successfully complete meaningful work, the titles will follow. As cliché as it sounds, the way Bilawal shares his personal story to demonstrate this is truly inspiring. 2. True success requires cultivating high agency. 3. Do not let your title define your limits. For example, even as a junior software engineer, you can reach out to a VP. All it takes is a simple coffee chat invite. Start with 3D tools to exert fine grain control on the things that matter, and use generative AI to take it all the way. This is the way. Physical AI is here. What is it, what makes it a 10^14 $ opportunity, and most importantly, how will it help all of us live our best lives. Spatial computing and XR play key roles. Check out this terrific analysis by Bilawal Sidhu on Physical AI, deconstructed.  One form Physical AI will take is the humanoid robot. We're excited to support breathing life into your very own C-3PO with the work we're doing on Isaac GR00T. Learn more here: https://lnkd.in/ekPm-87s