Samet KILICTAS’ Post

🤖📝 𝐂𝐨𝐝𝐞𝐂𝐨𝐦𝐩𝐨𝐬𝐞 - 𝐌𝐮𝐥𝐭𝐢-𝐥𝐢𝐧𝐞 𝐀𝐈-𝐚𝐬𝐬𝐢𝐬𝐭𝐞𝐝 𝐂𝐨𝐝𝐞 𝐀𝐮𝐭𝐡𝐨𝐫𝐢𝐧𝐠 If you are in software development & engineering then this is of interest. Gives insights how Meta uses AI and went from single to multi-line suggestions in their software engineering process showing some very positive results. 🛠️ CodeCompose is an AI-assisted code authoring tool powered by large language models (LLMs) that provides inline suggestions to 10's of thousands of developers at Meta. ⏱️ Speeding up multi-line suggestion generation by 2.5x. 💬 Experiments on tens of thousands of engineers reveal that multi-line suggestions significantly improve user experience, with 42% of total characters accepted and keystrokes saved nearly doubling. 🌐 Multi-line CodeCompose has been rolled out to all engineers at Meta, with less than 1% opting out. Abs: https://lnkd.in/g7deSheD

Here's a cute message by Meta AI for all of you coding warriors feeling overwhelmed by the next big challenge in your journey. "Hey there, coding warrior! Don't let bugs or errors intimidate you. Channel your inner Giga Chad - fearless, confident, and unstoppable. Remember, every line of code is a conquest, every debug a victory. Embrace challenges, learn from failures, and rise above. You got this! Giga Chad mindset tips: 1. Own your code. 2. Fear less, iterate more. 3. Debug like a boss. Unleash your coding beast. Crush those bugs. Build something epic!" Honestly, so wholesome ✨ #meta_ai #GenAI #programming #coding

Our forms are slow and glitchy. We're actively fixing it. Engineering problems compound for a living AI-native product. Usually the AI products that've gone big are exported out— voice clips, videos, graphics, presentations, copies etc. The engineering challenge there is very different than a product which gets its value from real-time generation even-post publishing. And Forms are also the primary interface a user has with a brand to communicate what they're thinking and feeling. A form failing is an equivalent of a brand saying don't talk to me, I don't care. Building this critical AI-native artefact is a challenging problem Metaforms is solving. It might take us some time, but we'll get there. Btw, we're also hiring for Full Stack, AI, Product and Design roles. https://lnkd.in/eBTZ8h4w

Meta Llama 3 came out less than a month ago, and has unlocked so much for builders! Nipun and I hacked away last weekend at the Meta hackathon, shipped all night overlooking the ships 🚢 at Embarcadero, SF! Built out a Zoom-like app that lets you choose any real/fictional character and talk with them, in real time, in their voice! They say, you're the average of the 5 people around you. The idea was to let you curate your Top 5, and to learn & grow by simply talking with your idols! It's wild that the tooling for this exists, and you can spin up something like this in under 24 hours. The stack we used: - Groq + Llama-3 for text completion >> INSANELY FAST! - ElevenLabs for voice gen (just needs one minute of audio as training!) - OctoAI for image gen - Whisper(faster-whisper) on local for transcription - Streamlit for the User Interface Sharing a video demo of the product in the comments - if you watch it, you'll realize how latency of 1-2 seconds still makes it feel a bit unnatural. Was blown away to see OpenAI then turn up the next morning, and roll out GPT-4O that breaks through that uncanny valley! Such a beautiful time to be building in 🚀

ML in academia vs industry I didn't have a typical path to big tech. My undergrad was in biomedical engineering, my master's in information and computer engineering and my PhD in statistics and computational biology. In my third year of PhD I did a software engineering internship at Bloomberg and joined the Search & NLP group there as soon as I submitted my thesis. I've now been at Meta for the last five years, working on recommender systems and ads delivery optimization. And I've never looked back. I always knew academia wasn't right for me. I wanted to solve real-life problems and build things. Unashamedly, I was also interested in making money and I did not see a way of making that happen in academia. Lastly, I liked to code, more than reading or writing papers. In my head, that’s what I would do *all the time* as a software/ML engineer in industry (plot twist coming up) while being paid a lot more. What was there not to like? The transition was a bit of culture shock. Back in academia I was used to working on my own and being my own bottleneck. I got to do *deep* work for hours/days on end.  That is rarely the case in industry. It’s still of course about solving difficult technical problems, but it’s also about navigating complex human dynamics, figuring out how to work in a team, how to align with others with different agendas and how to manage resources, timelines and competing priorities. The problem complexity is also different. In industry, I rarely *have* to read papers because I’ll usually know how to solve a problem from a modeling perspective. The complexity is about how to get our production systems to do that ‘simple’ thing we need. It’s about the complex interaction of components that are maintained and developed by thousands of engineers serving billions of users at a time. It is about your experiment getting corrupted by a team on the other side of the world making a change to a service you know nothing about. Nothing in academia prepares you for this challenge. And it's pretty fun. If you are interested in making the jump from academia to industry check out this job opening at Meta!

Meta open-sourced the code of their impressive SAM demo! I'm Milan this week for #ECCV2024, a big computer vision conference. It allows for meeting up with a lot of researchers, and discuss how to increase the discoverability of their work by releasing on Hugging Face. We encourage researchers to index their paper (by going to hf.co/papers/<your-arxiv-id>). Next: - models could be uploaded based on https://lnkd.in/emjZjgQJ - datasets could be uploaded using https://lnkd.in/eRC5z8Xj - a Gradio demo could be made based on https://lnkd.in/edJThZDv. I also met the incredible team behind Segment-Anything 2! Meta just released fresh SAM-2.1 checkpoints on Hugging Face: https://lnkd.in/ecUSuXMH. The authors just open-sourced the code (both front- and backend) of their super impressive demo at https://lnkd.in/egciaxH2. The code can be found at https://lnkd.in/eGFrQhrZ #segmentanything #sam2 #huggingface

You are what you eat. 10 brain foods to grow as an engineer: 0. How Meta Achieves 99.99999999% Cache Consistency (by Neo Kim): https://lnkd.in/d6723jdY 1. LeetCode was HARD until I Learned these 15 Patterns (by Ashish Pratap Singh): https://lnkd.in/dSibiWJW 2. Optimize Your Searches with Effective Database Indexing (by Nikki Siapno): https://lnkd.in/drp_-Gap 3. How a Single Line of Code Brought Down a Billion Dollar Rocket (by Dr Milan Milanović): https://lnkd.in/dZ93wqu6 4. Browser fingerprinting explained (+7 top techniques): https://lnkd.in/dzqRD8ps 5. Three Steps to Developing a Strong Network: https://lnkd.in/dAbDSqui I am packaging the other deep dives and trends in today's issue of Hungry Minds. Join 25,001+ software engineers from big tech to startups here: https://lnkd.in/dkJiiBnf What are you reading this week? #softwarengineering #systemdesign #programming

Meta CEO Mark Zuckerberg offered an apology during a US Senate hearing to families who shared stories of their children being harmed by bullying and exploitation on social media platforms owned by Meta, including Facebook and Instagram. This came after intense questioning from Senator Josh Hawley. Zuckerberg emphasized Meta's ongoing efforts to invest in making its platforms safer for children. follow Hancy Khalid Artificial intelligence Enginner for more tech info #ai #meta #childrens #news #artificialintelligence #machinelearning #technology #datascience #python #deeplearning #programming #tech #robotics #innovation #bigdata #coding #iot #computerscience #data #dataanalytics #business #engineering #robot #datascientist #art #software #automation #analytics #ml #pythonprogramming #programmer #digitaltransformation #developer

2 weeks ago I came across a post by Facebook which was written in 2022 on how they improved their cache consistency to 99.99999999 percent (10 nines) that means less than 1 out of 10 billion cache writes are inconsistent in meta after 5 minutes. I have tried to write a detailed blog on it and also tried to reproduce an actual bug which they solved during this process. I have implemented the code in python on my github https://lnkd.in/g3KmMat7. This blog was originally written by Lu Pan and I have tried to write it in my words and what I understood out of it. References : https://lnkd.in/ggQtPdqb https://lnkd.in/gWk59y-g #systemdesign #coding #softwareengineering #meta #facebook #caching

🚀 Exciting Update Alert: Journey, Skills, Projects, and Milestones with Meta Scifor Technologies ! 🌟 I'm thrilled to share a comprehensive update on my journey, skills, projects, and milestones at Meta Scifor Technologies With a commitment to accuracy and completeness, I've revamped my LinkedIn profile to reflect the latest developments in my professional journey with this innovative organization. Take a look and let's connect if you resonate with any of the shared experiences. 🛠️ Experience Highlights: During my tenure at Meta Scifor Technologies Technology, I've had the privilege of diving into diverse projects and acquiring invaluable skills, particularly in the realm of Data Science. From honing my expertise in data analysis, machine learning, and predictive modeling to navigating intricate challenges, each experience has been a stepping stone towards growth and excellence. 💼 Projects Completed: Reflecting on my journey at Scifor, I'm proud to have contributed to numerous impactful projects. Notably, I led the development of the ChatPDF app, revolutionizing how users interact with PDF documents seamlessly. Additionally, I spearheaded the implementation of Handwritten Digit Recognition, showcasing the potential of machine learning in pattern recognition tasks. 🏆 Milestones Achieved: Every milestone achieved at Meta Scifor Technologies has been a testament to dedication, perseverance, and teamwork. From successfully launching the ChatPDF app to achieving state-of-the-art accuracy in Handwritten Digit Recognition, each accomplishment has reinforced my belief in the power of data-driven solutions and collaboration. 🙏 Special Thanks: A heartfelt thank you to my trainer, Urooj Khan, whose guidance and support have been instrumental in shaping my journey and honing my skills. Your mentorship has been invaluable, and I'm grateful for the knowledge and wisdom you've imparted. Thank you for being part of this incredible journey with Scifor Technology. Here's to new beginnings and endless possibilities! Check out the projects on my GitHub repository:https://lnkd.in/g7DZfERk #ProfessionalJourney #DataScience #MachineLearning #Projects #Milestones #MetaSciforTechnology #LinkedInUpdate #Gratitude