Maitri Parikh

My team at Meta (In-Feed Recommendations) is hiring a number of ML engineers in the US We are building one of the most advanced recommendation engine in the industry to power discovery experiences on Facebook newsfeed with a diverse set of public content (Reels, videos, photos, web links, conversations, etc) available on our platform. You can make a big difference by working on a wide range of projects, including but not limited to user/content interests understanding, next-gen recommendation systems and modeling technologies, new social discovery experiences to help create communities of like-minded people. Check out the following links to learn more: The AI Behind Unconnected Content Recommendations on Facebook and Instagram https://lnkd.in/gZ7XV_am Meta Rethinks the Philosophy of the Facebook Feed https://lnkd.in/ggJzk3g3 If interested, please DM me. ==================== Team / Role Summary We are part of the FB Feed Relevance Org, and our team’s mission is to create joyful discoveries and meaningful connections via personalized recommendations on people’s home feed with deep understanding of their interests. We are responsible for building the E2E recommendation system to power all the interest-based public content discoveries across FB Feed. Why is This Impactful IFR is a central product in Facebook’s long term vision, Discovery Engine. Each day we are sending more than 50B posts to our users. From Mark Zuckerberg's Q2 earnings call: "AI-recommended content from accounts you don't follow is now the fastest growing category of content on Facebook's feed. Since introducing these recommendations, they’ve driven a 7% increase in overall time spent on the platform. This improves the experience because you can now discover things you might not have otherwise followed or come across." Your Role As an ML Engineer, you will help build one of the world’s largest recommendation engine, implement state of the art ML algorithms and modeling techniques, directly influence billions of users’ News Feed Experience. You will also learn how to build an ultra large scale Recommender System from end to end and grow your career as a domain expert, and lead our efforts to build emerging new experiences on the Newsfeed Recommendations platform.

63

1 Comment

I reached 17 years at Google two weeks ago. Google has changed quite a bit over the years, though I won't get into that now. Sometimes, though, it's not the big things you've landed, but the smaller things that unexpectedly turned into bigger things. I frequently reflect on: - Android lint testing for Google apps - Fake "Save" button in Google Docs - The algorithm for Google Calendar's event layout - A couple of random test frameworks I created along the way, e.g. one that repeatedly prevented new crashes in Android OS itself - Some novel algorithms for detecting production issues in Android - Helping land XSS-free autoescaping for one of our most common HTML templating languages Many of these were things I started on a whim because I thought they might be valuable. Some of them I landed myself, but some were just started and took on lives of their own. Of course, I tried a lot of other things that didn't work, but it's important to try and experiment even when not everything works. What small things bring _you_ disproportionate pride?

290

19 Comments

𝗙𝗿𝗶𝗱𝗮𝘆, 𝟮𝟰 𝗠𝗮𝗿𝗰𝗵 𝟮𝟬𝟮𝟯: Was crying on call with my aunt, saying that I've been working on maltiplai since September 2022 but have no idea where this is going. 𝗦𝗮𝘁𝘂𝗿𝗱𝗮𝘆, 𝟮𝟱 𝗠𝗮𝗿𝗰𝗵 𝟮𝟬𝟮𝟯, 𝟭𝟬:𝟬𝟬 𝗔𝗠: Back to working on maltiplai when a voice inside me told me look for courses at MIT(still don't know where that voice came from). I remember almost paying for an online course in Neural Networks, but I stopped and ended up giving an interview for an onsite Venture Advancement Program ​MIT Bootcamps 😍​ 𝗠𝗼𝗻𝗱𝗮𝘆, 𝟮𝟳 𝗠𝗮𝗿𝗰𝗵 𝟮𝟬𝟮𝟯: I got "the email"! It stated "It will be a life-changing experience"(I really needed one!) 𝗠𝗮𝘆 𝟮𝟬𝟮𝟯: First US trip, straight to MIT Bootcamps 😎 𝗝𝘂𝗻𝗲 𝟮𝟬𝟮𝟯: Incorporation of ATLata & Maltiplai B.V. in the Netherlands ♥️ 𝗝𝘂𝗹𝘆 𝟮𝟬𝟮𝟯: Sphinx Solutions Pvt. Ltd. onboarded for software development 𝗝𝗮𝗻𝘂𝗮𝗿𝘆 𝟮𝟬𝟮𝟰: maltiplai went live on iOS & Android in 60 countries! 𝗙𝗲𝗯𝗿𝘂𝗮𝗿𝘆-𝗔𝗽𝗿𝗶𝗹 𝟮𝟬𝟮𝟰: maltiplai collaborated with Amity University, Miranda House, Shri Ram College of Commerce (SRCC), Delhi, Hindu College, University of Delhi, Aryabhatta College, University Of Delhi, Sri Venkateswara College, Delhi University, Shaheed Sukhdev College Of Business Studies, NSUT Delhi, Delhi Technological University (DTU), Delhi College Of Arts and Commerce and University School of Automation and Robotics (USAR) to spread the word that we are here! 𝗠𝗮𝘆 𝟮𝟬𝟮𝟰: Got a chance to share advice with the next group of innovators — just a year ago, I was sitting in the same room ♾️ I haven't done this alone; I'm 𝗯𝗹𝗲𝘀𝘀𝗲𝗱 𝘄𝗶𝘁𝗵 𝗽𝗲𝗼𝗽𝗹𝗲 𝗼𝗳 𝗵𝗶𝗴𝗵 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲. Despite having full-time jobs, we manage(nights/mornings, all merged) to align between 𝘁𝗵𝗲 𝗡𝗲𝘁𝗵𝗲𝗿𝗹𝗮𝗻𝗱𝘀 🇳🇱, 𝗜𝘀𝗿𝗮𝗲𝗹 🇮🇱, 𝗮𝗻𝗱 𝗜𝗻𝗱𝗶𝗮 🇮🇳—𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮𝗻 𝗶𝗻𝘁𝗲𝗿𝗻𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗳𝗿𝗼𝗺 𝗱𝗮𝘆 𝗼𝗻𝗲. 𝗕𝗼𝗼𝘁𝘀𝘁𝗿𝗮𝗽𝗽𝗲𝗱, after 5 months and 14,000+ downloads, we're preparing to deploy an improved product and even launch our revenue channels. We might not get it right the first time, but we are here for the long run, 𝗳𝗼𝗿 𝘄𝗲 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗶𝗻𝗳𝗶𝗻𝗶𝘁𝗲 𝗽𝗹𝗮𝘆𝗲𝗿𝘀 😎 #CreatorEconomy #TheFutureIsHere

58

14 Comments

Vijay Kulkarni Harish Nagu Sana and I wrote a Medium blog titled "Stateful Feature Backfill on Apache Beam." We have learned a lot about Apache Beam, including checkpointing, savepointing, state management, and much more during the project. Check it out! #apachebeam #intuit #intuitengineering #featuremanagement #intuitAi https://lnkd.in/gVwxaeby

37

2 Comments

Excited to share our blog on a stateful historical backfill feature using Apache Beam. Our team developed a unified pipeline for processing historical data and then to real-time. #apacheBeam #streamProcessing #featureManagement #intuit #intuitAI #apacheFlink

17

2 Comments

🎬 AI-Generated Movie Trailer, casting Yourself: Sounds fun? 🤖 I recreated scenes from the Bollywood hit 'War' - starring myself! 🕺 Here is the instagram reel: https://lnkd.in/gwkqdw3R What if I told you this movie trailer was created entirely by AI? 🤯 This is not a deepfake where we swap faces but each scene AI regenerates from the learning data sets (my images + movie images). This technique is called Fine Tuning! 🛠 Tools 🛠 1. Flux Dev for fine tuning the model to train on my images and movie scenes and generate new movie images from scratch 2. RunwayML Gen3 for image to video 3. CapCut for video editing and stitching things together The best part? I used minimal prompts and basic training data. Yet the results are mind-blowing! 🚀 The future of filmmaking is here - are you ready to be a part of it? 🎥✨ 🤔 Curious about how I did it? 🤔 Here is the medium blog for step-by-step guide: https://lnkd.in/gr6z_CqJ 🔥 Follow me & stay tuned for more AI related videos! 🔥 Credits: 100xEngineers Sridev Ramesh Siddhant Goswami for an amazing masterclass to teach how to make such videos & more, YRF-Yash Raj Films for movie artifacts available online. #AIFilmmaking #GenerativeAI #ai #0to100xEngineer #yrf #finetuning #hritikroshan #warhindimovie #flux1dev #fluxdev #fluxlora #runwayml #capcut

54

9 Comments

🚀Hi Folks, I'm Taking on the Product #Teardown_Challenge along with Arindam Goswami! 🚀 As part of an exciting challenge by Product Space, I’m stepping into the role of Senior VP of Product at #Zepto Quick Commerce! Zepto’s already crushing it with 10-minute deliveries across 30 cities in India, but there’s so much more potential to unlock! 💡 Here’s the mission: ➡️ Increase Zepto’s Average Order Value (AOV) to ₹600 🛒 ➡️ Boost product discovery and engagement to encourage bigger, more frequent orders 🔍 ➡️ Elevate the purchase experience to keep customers coming back for more 🔄 And I need YOUR help to make it happen! 🙌 🎯 Your insights on quick commerce platforms like Zepto, Blinkit, and Swiggy Instamart are key to shaping these strategies. I’ve put together a short survey, and your feedback will be crucial in tackling this challenge! 💥 Thanks for your support—let’s crack this together! 🙌 Link to the form - https://lnkd.in/g5v2y4wR #ProductTeardown #QuickCommerce #ProductManagement #ProductSpace

9

2 Comments

Amazon-to-Google-to-Amazon #a2g2a thought of the day: the way Amazon thinks about its Principal Engineers is unique - and I love it. “I always thought of you (and my other Principal Engineers) as Amazon employees that were on loan to me.” Yesterday I was catching up with my good friend and ex-manager Dan Sommerfield. Dan promoted me to Principal Engineer back in 2014 and was my manager for a number of years. As we were laughing about some of our shared past, Dan said that and it stuck with me. A few weeks ago I posted in my a2g2a series about Swimming Lanes. I’ve also posted in the past about the Amazon Principal Engineering community (see links to both of those in my comments). Today’s reflections combine those. Amazon Principal Engineers tend to have a healthy disregard for organizational and role boundaries. And Amazon’s DNA gives them the freedom to do so. At these big faang companies, for better or for worse, most orgs tend to optimize for themselves. It makes sense in terms of business needs: they need to move fast and expand their business. But this is often how we end up with duplicative work siloed usually at Director level. I’ve seen, time and time again: Principal Engineers look across organizations and connect dots that others don’t. The Amazon Principal Engineering community is a giant organic mesh. It worked incredibly well as a big mesh when I became a PE in 2014 and we were about 250 PEs in the entire company, but I’m continuously amazed that it still works in 2024 at today’s scale. This is a testament to the culture that some of our founding figures set decades ago- and our Distinguished Engineers continue to carry the torch today. It’s not that this doesn’t happen in other companies; it did happen at Google and Microsoft too. But it’s so intrinsically embedded in the DNA of an Amazon Principal Engineer that it’s a lot more natural and expected here than anywhere else where I’ve worked or considered working.

594

24 Comments

Here is a quick look at how to achieve high-quality semantic search. Lot of exciting work happening in this space. Solving real world problems with AI. #googlecloud #vertex #AI #google

31

Our SIGMOD '24 paper explores MemoryDB's architecture, which delivers in-memory performance and strong consistency at scale. A key innovation is the decoupling of durability and replication from the in-memory database engine, which allows scaling compute and storage independently based on customer workload needs. Customer data integrity, security, and availability are paramount for MemoryDB. The paper dives deep into innovations that the team built to make that possible (a personal favorite is off-box snapshotting which allows us to create consistent database snapshots on separate clusters to avoid impacting live customer traffic and preventing resource contention). We also share hardened operational experiences from running a distributed database at cloud scale. #aws #memorydb #inmemory #databases #durability #sigmod

91

1 Comment

"Attention Is All You Need" This is the whitepaper which made GPT possible. This was a revolutionary paper which enabled AI and is the launchpad for all AI development that you see now. I have started reading it. If you have time, read it. Google about it. This paper is freely available to read in html / pdf etc format. #ai #testing #automation #softwareengineering

42

7 Comments

Countdown to OpenAI's Spring update on Monday at 10am PT (i.e. right now) Live Spoilers from Mira Murati: - ChatGPT Desktop - ChatGPT 4o available to free users -- smarter and faster voice, text, vision -- stored conversational memory -- improved quality in 50 languages (97% of world population) -- API 2x faster 2x cheaper Demo: - Voice conversation where the AI gives feedback on breathing pace - AI is telling a bedtime story; user interrupts the AI speaking and asks for dramatic or robotic tone, AI gets on-board in in real-time and speaks dramatically or robotically - User writes a math equation on a piece of paper and asks AI to help along with solving it. The AI watches the user work through it and gives hints. - User writes "I <3 chatGPT" and AI responds as if touched "emotionally". - User pastes complex code and asks AI to describe what the code does in one sentence. AI knows what the code is doing and can answer specific questions. - User executes code which produces a weather diagram. GPT reasons about weather events and describes what happened by "watching" the diagram. - Mira and Mike role play as if one only speaks Italian and the other, only English. GPT acts as the live voice translator. - User asks GPT to look at his live face and read emotions, GPT says "happy with a touch of excitement". User says I am excited, I'm describing how amazing you are. GPT says with emotional intonation "Oh stop it!"

3

I am excited to write this blog together with (Kevin) Huan-Ping Su from Union. - Anyscale is the best place to run Ray workloads, and Ray is the go-to solution for ML and platform engineers to manage the end-to-end lifecycle of ML workloads as a unified computing engine. - Union.ai is the best place to orchestrate Flyte pipelines, and Flyte is a great solution to address gaps within ML pipelines, such as managing task lifecycles for infra engineers and accessing data from data warehouses. This blog will dive into a RAG example to show the perfect marriage between Anyscale and Union. Sign up for Ray Summit (raysummit.anyscale.com) to learn more about Ray!

61

How do you make AI feel human? 🤖😊 It starts with great design. Just wrapped up the Becoming an AI Product Designer course with Maheen Sohail, and I must say—for a PM who loves the intersection of taste and technology, it was a real treat. Maheen, with her experience designing cutting-edge AI products at Meta, took us on a journey through the essential patterns shaping the new generation of AI-powered experiences. One of my favorite sections? AI Design Patterns. Here’s a glimpse of what we explored: ▪️Education and NUX: How Duolingo and DOT guide users effectively through onboarding ▪️Contextual UI: Inline actions and side panels that keep the flow intuitive and in control ▪️Cold Start Solutions: From icebreakers to guided inputs, reducing the fear of that empty canvas ▪️Continued Guidance: Keeping engagement high with suggested prompts and ongoing assistance ▪️Personalization: Memory, suggestions, and context that make it feel like the product knows you ▪️Experimentation: Preview, undo, retrace—the keys to empowering users with AI ▪️Feedback Mechanisms: Thumbs up/down, options to regenerate, and continuous refinement ▪️Providing Controls: Source verification, image editing—giving power back to users ▪️Transparency: Labels, watermarks, and clear data usage to ensure trust ▪️Voice & Streaming: Exploring new dimensions in how we interact with products Maheen brought each of these design patterns to life with real-world examples and additional resources. The class experimented with designing new AI patterns in their final AI showcase projects—it was really eye-opening to see how powerful these new patterns can enrich the user experience! It truly feels like we’re at the dawn of something exciting in product design powered by AI.🌟 If you’re a PM or designer stepping into AI, I highly recommend the course. In just three weeks, we covered: - AI & ML Fundamentals and their impact on Design - Design challenges for AI products - Responsible AI, Design Principles & Ethics - The Design process for AI products - Skills required in AI Design And a special shoutout: there's a 𝟮𝟱% 𝗱𝗶𝘀𝗰𝗼𝘂𝗻𝘁 for the next cohort! Use code 𝗦𝗛𝗬𝗩𝗘𝗘 to unlock it. Discount ending TODAY!!! (After today it will go back to $100 off) 👉Learn more: https://bit.ly/3UEse5o  Let’s build the future together. 🚀 #ProductManagement #ProductDesign #AI

42

9 Comments

AI LLMs cannot know everything, so function calling agents now have the ability to call external APIs and get data/do stuff. I wrote a simple llamaindex function calling agent to get the latitude and longitude for a given city. I will soon use it to get weather data. Such agents can order food, send a reminder email, or read your appointment calendar ( if there is a REST API).

3

1 Comment

What’s the Real Context Size of Your Long-Context LLMs? Picking up from my last post (https://lnkd.in/gYtgs_DT), where I talked about the pros and cons of long-context LLMs in Retrieval-Augmented Generation. This NVIDIA research takes a closer look at how these models perform at their limits, with insightful findings that reveal some non-obvious challenges. The authors tested 17 language models, including well-known ones like Gemini 1.5, GPT-4, and Llama 3.1. Turns out, despite many models claim to handle huge contexts, very few actually keep up their performance as context length grows. This might come as a surprise if you're familiar with traditional benchmarks like "needle-in-a-haystack" and passkey retrieval, where most models achieve near-perfect accuracy. While those tests are useful, the new RULER benchmark pushes LLMs with more complex tasks, such as multi-hop tracing and aggregating information across long contexts, ultimately revealing their true limitations and areas for improvement. As language models grow in size and context capacity, it's easy to get excited about their new capabilities and forget the trade-offs. Just because a model can handle massive contexts doesn’t mean more is always better — often, keeping it small and simple still wins the day. I'll drop the link to the paper in the comments. #MachineLearning #ArtificialIntelligence #AI #LLM #GenAI #NeuralNetworks

58

3 Comments

Fantastic news today on the general availability of Amazon Q. Amazon Q is the most capable generative AI-powered assistant for accelerating software development and leveraging companies’ internal data. It eliminates tedious work for developers and employees across organizations. #AmazonQ #GenerativeAI

15

My teams' feedback on trainium chips from Amazon Web Services (AWS) - AWS's version of pytorch makes it such that we have to port model training done in the past for NVIDIA GPUs - that's a non-starter. If AWS wants more folks to use their chips for training/inference they should be first byte-for-byte compatible with pytorch used widely across the industry to train on NVIDIA. We would be okay, with some magic underneath the covers but the interface to pytorch can't change one bit.

19

5 Comments