VIAO.AI的动态

?? Lessons from the Journey Looking back, the transition from academic learning to real-world applications taught me: 1?? Practical experience is as crucial as theoretical knowledge. 2?? Collaboration across disciplines (data and development) creates impactful solutions. 3?? Lifelong learning is the key to staying relevant. Now, I’m focusing on mastering advanced topics like machine learning pipelines, API integrations, and creating more scalable solutions. My advice to fellow students: Start small, think big, and never stop building. ?? #GrowthJourney #BTechToPro #DataAndDev

Hello! ?? I'm excited to share my recent collaboration with Lightning AI in developing new Studio templates! ??? They have created a cloud-based, reproducible data science environment designed to streamline workflows and boost productivity. In the modern data science landscape, MLOps plays a critical role in creating reliable, production-ready pipelines. Ensuring consistency, reproducibility, and efficient data engineering is essential for the success of any ML project. Why MLOps Matters in Data Science: ?? Reproducibility: Ensures that your experiments can be reliably reproduced. ?? Scalability: Facilitates scaling models to handle larger datasets. ?? Efficiency: Streamlines the workflow, saving time and resources. ??? Reliability: Creates robust and reliable ML pipelines. This corresponds directly with features of the Lightning AI Studios: ?? Cloud-based Environment: Access and manage your projects from anywhere. ?? Reproducible Pipelines: Easily recreate experiments and results. ?? Scalable Solutions: Handle projects of any size with ease. ??? Integration with Leading Tools: Works seamlessly with tools like Tensorboard or Streamlit. I’ve been invited to collaborate in creating useful, execution-ready studios using the Kedro framework, an open-source library for orchestrating and building reliable ML pipelines. If you want to learn how Kedro can help you in MLOps management, check out my Lightning AI Studio. ?????? https://lnkd.in/dYZqSagc #MLOps #DataScience #Kedro #LightningAI #DataEngineering #Reproducibility #MachineLearning #AI

?? Dive into My GitHub: Where Data Meets Code! ?? I've been building a collection of projects that combine data science and machine learning to tackle practical challenges. My GitHub is a place where I experiment, learn, and share the results openly with the community. If you're curious about: Predictive modeling techniques ?? Machine learning pipelines ?? Real-world applications of AI ?? …then take a look and feel free to reach out—I’d love to connect with like-minded professionals! ?? GitHub: https://lnkd.in/etZVAnHq #DataScience #MachineLearning #GitHub #AI #OpenSource #Coding #TechCommunity

My GitHub Wrapped for 2024 is in! I wrote close to a million lines of code this year. Here are some of the highlights of what I built for AllMind AI: ? Built a NoSQL database that can query over a billion rows in under 4 seconds. ? Trained a small, in-house LLM with our own embedding model. ? Put together an MoE LLM pipeline using fine-tuned open-source LLMs. ? Created a live websocket connection for real-time AI/ML analysis of stock trades. ? Built a system to spot institutional dark pool trades. ? Handled tens of thousands of user queries on AllMind AI. ? Developed a real-time detector for bias in earnings calls and the resulting market impact. This year, I started January with the idea of building a company which later turned into AllMind AI, and in our first year, we've hit some major goals: we've got tens of thousands of users, built a ton of cool stuff, added new people to the team, and even presented to some great groups like RBC and given talks at several universities! It's been a blast, and I can't wait to see what's in store for next year!

Hey, I recently completed the "Build Your Machine Learning Pipeline with Kubeflow" webinar, presented by Andreea Munteanu (AI/ML Product Manager) and Kimonas Sotirchos (Kubeflow Software Engineer). Over the course, I gained valuable insights into building robust machine learning pipelines using Kubeflow. This session was a great opportunity to deepen my understanding of deploying and scaling ML workflows, a crucial skill in the ever-evolving tech landscape. Thank you, BrightTALK , for hosting this insightful webinar! Here's to more learning and growth on my journey in ML and software engineering. #MachineLearning #Kubeflow #Ubuntu

?? Just wrapped up Module 5 of the MLOps Zoomcamp with Emeli Dral at DataTalksClub ?? Explored essential strategies for monitoring machine learning models: Introduction to ML Monitoring: Learned why it's crucial to monitor ML models throughout their lifecycle. Environment Setup: Set up environments to ensure consistent monitoring practices. Reference and Model Preparation: Established baseline metrics and prepped models for effective monitoring. Evidently Metrics Calculation: Calculated key metrics using Evidently to assess model performance. Evidently Monitoring Dashboard: Built interactive dashboards to visualize and track model metrics. Dummy Monitoring: Implemented initial monitoring techniques to check model health. Data Quality Monitoring: Ensured data quality and integrity throughout model deployment. Save Grafana Dashboard: Saved Grafana dashboards for ongoing monitoring. Debugging with Test Suites and Reports: Used test suites and reports to debug and optimize model performance. ??? Module 5 was challenging yet incredibly insightful, equipping me with crucial skills to maintain model health and performance. ?? Huge thanks to Emeli Dral and Alexey Grigorev for this hands-on learning experience in MLOps! Excited to apply these skills in real-world projects. ?? Check out the GitHub for more details on this project and others: [https://lnkd.in/ekMjU3nu] ?? Also, explore my GitHub to see my journey: [https://lnkd.in/eNCd-Bye] #MLOps #Monitoring #MachineLearning #DataScience #DataTalksClub

We’ve got some exciting news—Raggenie just launched on Product Hunt, and it’s open-source! Here’s a summary of what it can do: 1. Structured data: Connect to databases, query, and visualize your data in real-time—essentially, chat with your databases & data lakes 2. Unstructured data: Ask questions and get answers from documents, websites, and more. 3. Agentic workflows (coming soon): Automate tasks like webhooks and database create/update operations. 4. Low-code: No technical expertise is required to get started! Raggenie can be used for personal projects, shared with others, or integrated into your apps moreover we are open for contributions to the project Check it out here: https://lnkd.in/ghPKDQhj #opensource #raggenie #producthunt #launch

- My final project for the LLM Zoomcamp course! ?? I've developed a RAG application that leverages large language models to provide personalized travel recommendations and insights. The project includes a Flask API, data ingestion, and monitoring dashboards etc. ?? Key Features: Personalized travel recommendations User feedback tracking Real-time monitoring dashboard Check it out on GitHub: https://lnkd.in/gwbmCQYN I appreciate any feedback or suggestions for improvement! #LLMZoomcamp #MachineLearning #OpenAI #Flask #GitHub #RAG #GenerativeAI #GenAI

?? Leveling Up with GitHub Foundations! I’m excited to share that I’ve completed the GitHub Foundations track on DataCamp, one of the best learning platforms I’ve come across! ??? Over 9 hours of hands-on learning, I’ve mastered essential GitHub skills like repository management, version control, and team collaboration. These skills are game-changers for developers and data professionals alike. ???? If you’re looking to upskill in tech and data, I can’t recommend DataCamp enough! ?? This is just the beginning—excited to apply these new skills and keep growing. Here’s to learning, innovating, and creating together! ?? #GitHub #DataCamp #LifelongLearning #TechSkills

?? Exciting Insights from Xebia's Latest GitHub Copilot Survey! Discover how GitHub Copilot is transforming software development! Developers are finding that this #AI tool significantly saves time on basic coding tasks and enhances job satisfaction. But what about code quality? Our survey reveals the real impact of Copilot on productivity and quality. Dive into our findings and see how developers balance these aspects. Read more: https://lnkd.in/gJ_dpuXi #GitHubCopilot #AI #SoftwareDevelopment #Xebia #TechInnovation