Open Source Tools for Machine Learning Projects

5 hidden gem ML resources that actually get you hired. These are NOT your average MOOCs, they’re tools made for engineers like you to: • Solidify the math • Improve system design fluency, and  • Enable production-grade ML deployment 𝟭. “𝗧𝗵𝗲 𝗛𝗶𝗱𝗱𝗲𝗻 𝗟𝗶𝗯𝗿𝗮𝗿𝘆” 𝗯𝘆 𝗦𝗮𝘁𝘃𝗶𝗸 𝗦𝗵𝗿𝗶𝘃𝗮𝘀𝘁𝗮𝘃𝗮 A precision-picked set of 18 deep-dive posts, from reverse-mode autodiff and einsum to LLaMa internals and diffusion models, that finally make the core mechanics of ML click. https://lnkd.in/eKzCeR9f 𝟮. 𝗚𝗶𝘁𝗛𝘂𝗯: 𝗔𝘄𝗲𝘀𝗼𝗺𝗲 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿 A roadmap curated by the community, segmented into essential, advanced, and expert readings, designed to guide experienced engineers through what to study next, based on your level. https://lnkd.in/eUfcs7km 𝟯. 𝗗𝗶𝘃𝗲 𝗶𝗻𝘁𝗼 𝗗𝗲𝗲𝗽 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 (𝗗𝟮𝗟) An open-source, Jupyter-notebook-based textbook blending rigorous theory, math, and runnable code, built for deep technical mastery beyond tutorials. https://lnkd.in/eNwrAQP6 𝟰. 𝗔𝘄𝗲𝘀𝗼𝗺𝗲 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 A meticulously curated GitHub list of tools and libraries for deploying, scaling, monitoring, and securing ML in production, critical for engineers building real systems. https://lnkd.in/e78ayd3E 𝟱. 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 𝗗𝗲𝘀𝗶𝗴𝗻 – 𝗖𝗵𝗶𝗽 𝗛𝘂𝘆𝗲𝗻 A focused booklet (and companion to an O’Reilly book) that walks you through designing ML systems holistically, from project setup and data pipelines to modeling, serving, and iterative maintenance. One of the best resources for ML engineers building production systems. https://lnkd.in/epmtCmUw Build a project that has: • A solution to a business problem, • Clean, version-controlled code (Git + CI/CD), • A feature pipeline (maybe with a feature store), • Deployment with monitoring and feedback loops PS: If you're looking for your next $100K+ data job in the US or Canada, DM me the word "INFO" and I'll guide you to create your custom roadmap.

The Future of AI is Open-Source! 10 years ago when I started in ML, building out end-to-end ML applications would take you months, to say the least, but in 2025, going from idea to MVP to production happens in weeks, if not days. One of the biggest changes I am observing is "free access to the best tech", which is making the ML application development faster. You don't need to be working in the best-tech company to have access to these, now it is available to everyone, thanks to the open-source community!   I love this visual of the open-source AI stack by ByteByteGo. It lays out the tools/frameworks you can use (for free) and build these AI applications right on your laptop. If you are an AI engineer getting started, checkout the following tools: ↳ Frontend Technologies : Next.js, Vercel, Streamlit ↳ Embeddings and RAG Libraries : Nomic, Jina AI, Cognito, and LLMAware ↳ Backend and Model Access : FastAPI, LangChain, Netflix Metaflow, Ollama, Hugging Face ↳ Data and Retrieval : Postgres, Milvus, Weaviate, PGvector, FAISS ↳ Large Language Models: llama models, Qwen models, Gemma models, Phi models, DeepSeek models, Falcon models ↳ Vision Language Models: VisionLLM v2, Falcon 2 VLM, Qwen-VL Series, PaliGemma ↳ Speech-to-text & Text-to-speech models: OpenAI Whisper, Wav2Vec, DeepSpeech, Tacotron 2, Kokoro TTS, Spark-TTS, Fish Speech v1.5, StyleTTS (I added more models missing in the infographic) Plus, I would recommend checking out the following tools as well: ↳ Agent frameworks: CrewAI, AutoGen, SuperAGI, LangGraph ↳ Model Optimization & Deployment: vLLM, TensorRT, and LoRA methods for model fine-tuning PS: I had shared some ideas about portfolio projects you can build, in an earlier post, so if you are curious about that, check out my past post. Happy Learning 🚀  There is nothing stopping you to start building on your idea! ----------- If you found this useful, please do share it with your network ♻️ Follow me (Aishwarya Srinivasan) for more AI educational content and insights to help you stay up-to-date in the AI space :)

IBM 💙 Open Source Our AI platform, watsonx, is powered by a rich stack of open source technologies, enhancing AI workflows with transparency, responsibility, and enterprise readiness. Here's the list of key projects: 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 & 𝗩𝗮𝗹𝗶𝗱𝗮𝘁𝗶𝗼𝗻: - CodeFlare: Simplifies the scaling and management of distributed AI workloads by providing an easy-to-use interface for resource allocation, job submission, and workload management. - Ray / KubeRay: A framework for scaling distributed Python workloads. KubeRay integrates Ray with Kubernetes, enabling distributed AI tasks to run efficiently across clusters. - PyTorch: An open-source framework for deep learning model development, supporting both small and large distributed training, ideal for building AI models with over 10 billion parameters. - Kubeflow Training Operator: Orchestrates distributed training jobs across Kubernetes, supporting popular ML frameworks like PyTorch and TensorFlow for scalable AI model training. - Job Scheduler (Kueue/MCAD): Manages job scheduling and resource quotas, ensuring that distributed AI workloads are only started when sufficient resources are available. 𝗧𝘂𝗻𝗶𝗻𝗴 & 𝗜𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲: - KServe: A Kubernetes-based platform for serving machine learning models at scale, providing production-level model inference for frameworks. - fms-hf-tuning: A collection of recipes for fine-tuning Hugging Face models using PyTorch’s distributed APIs, optimized for performance and scalability. - vLLM: A fast and flexible library designed for serving LLMs in both batch and real-time scenarios. - TGIS (Text Generation Inference Server): IBM’s fork of Hugging Face’s TGI, optimized for serving LLMs with high performance. - PyTorch: Used for both training and inference, this is a core framework in watsonx. - Hugging Face libraries: Offers a rich collection of pre-trained models and datasets, to provide cutting-edge AI capabilities. - Kubernetes DRA/InstaSlice: DRA allows for dynamic resource allocation in Kubernetes clusters, while InstaSlice facilitates resource sharing, particularly for GPU-intensive AI tasks. 𝗔𝗜 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗟𝗶𝗳𝗲𝗰𝘆𝗰𝗹𝗲: - Kubeflow & Pipelines: Provides end-to-end orchestration for AI workflows, automating everything from data preprocessing to model deployment and monitoring. - Open Data Hub: A comprehensive platform of tools for the entire AI lifecycle, from model development to deployment. - InstructLab: A project for shaping LLMs, allowing developers to enhance model capabilities by contributing skills and knowledge. - Granite models: IBM’s open source LLMs, spanning various modalities and trained on high-quality data. We're committed to the future of Open Source and its impact on the AI community.

Most AI teams still don’t evaluate their models rigorously. Giskard is one of the leading open-source frameworks I’ve seen for fixing that. Giskard isn’t new, but it has evolved into a mature, production-ready testing toolkit that helps developers catch issues before they reach users. It supports both LLM agents (like RAG pipelines) and classic ML models, and automates the detection of: (1) Hallucinations and harmful generations (2) Prompt injection vulnerabilities (3) Robustness failures and edge cases You can wrap your model in a few lines of Python and run targeted scans that surface detailed test cases you’d otherwise miss. If you’re building with LLMs or ML models and not running structured tests yet, Giskard is the easiest place to start. GitHub repo https://lnkd.in/gTHen-Xu — This repo and 40+ curated open-source frameworks and libraries for AI agents builders in my recent post https://lnkd.in/g3fntJVc

If I was completely new to MLOps, this is the repo I’d start with. (from simple concepts to an end-to-end project) It doesn’t assume you’re already an ML expert — it walks you through the basics of ML processes first, then guides you step by step toward building and deploying a complete project. This repo is structured week by week, along with DevOps practices: Week 1 – Introduction & Experiment Tracking (MLflow) Week 2 – Data Versioning & Orchestration (DVC, Prefect) Week 3 – Orchestration continued (Prefect pipelines) Week 4 – Model Deployment (FastAPI, Docker, Kubernetes) Week 5 – Model Monitoring & Observability Week 6 – CI/CD for ML (GitHub Actions, unit/integration tests) Week 7 – Project: end-to-end ML system What I like most about this repo is how it connects the dots — from learning ML concepts → to running pipelines → to deploying and monitoring models in production. GitHub Repo with weekly modules: 🔗 https://lnkd.in/dH2SUPPK If you’ve been waiting to get hands-on with MLOps, you should definitely check this out! • • • I share bite-sized insights on Cloud, DevOps & MLOps (through my newsletter as well) — if this was useful, hit follow (Vishakha) and share it so others can learn too!