What awaits The AI Engineer in 2024
Local models, Small Models on the edge, podcast and a regular newsletter
We're one week into 2024 - the perfect time to reflect on what this community is all about, our goals for the year, and where AI might go next.
The AI Engineer's Essence
Recently, we received some constructive criticism about spotlighting OpenAI's Whisper, released in 2022. Ancient, they said! This comment prompts us to clarify our criteria for featuring software libraries and whether covering older ones is beneficial.
Our sole focus is empowering you to become the best AI engineer possible. Whether you're building AI apps for work or entrepreneurial ventures, seeking that dream job in the AI sector, or championing generative AI initiatives within your organization – our commitment is unwavering: you, the (nascent) AI engineer.
What We Embrace
Libraries that deliver results
Tools that streamline your tasks
Open-source software accessible to every developer
What Takes a Backseat
Mainstream media-covered AI news and discussions
Academic papers without practical applicability or easy-to-use implementations
Anything unverifiable without local installation or account creation
Our Pledge
We promise to promote sponsorships with clear labeling. We will never recommend you to try a software library that we would not use ourselves for our work.
Our "AI Library of the Day" series remains unsponsored. Each library is curated with a simple question in mind: "Will this help AI engineers excel?"
So, it's acceptable to delve into popular and not-so-recent libraries like Whisper or LangChain because they work. LangChain, despite the cool kids' dismissal, has been a crucial starting point for many embarking on their AI journey in 2023.
We view technology for what it is – neither endorsing the "killer robots" narrative nor claiming AGI is imminent.
We won't suggest that transformer architectures "understand" what they're doing, and we'll only cover content that aids your journey to becoming a better AI engineer.
What Awaits You in 2024?
Firstly, expect more diverse content in various formats. We're launching a weekly podcast and committing to a weekly newsletter, aiming for at least 50 issues this year.
Podcast Highlights
Our podcast features interviews with AI engineers and creators of AI products. Dive into discussions covering product and technological perspectives and operational and commercial aspects like business models and go-to-market strategies. We'll also explore content for Spanish-speaking audiences, providing translated transcripts and key takeaways.
While our online platforms focus on open-source libraries and operational discussions, the podcast and YouTube formats broadly discuss AI products and their creation from a helicopter perspective. We aim to avoid deep technical details on the podcast, recognizing that such discussions are better suited for visual formats.
Our 2024 Predictions
Open Source LLMs will rise but may not surpass OpenAI's offerings. However, they'll become user-friendly and powerful, potentially rivaling GPT-3.5. Cloud APIs still dominate agent-related tasks, but the closing gap in performance will be pushing the so-called "AI thin wrappers" and many applications toward open-source models.
Traditional databases like Redis or Elastic will be the go-to for storing vector embeddings. Vector databases might remain a playground for advanced users, while those new to AI app development will likely opt for pgvector, RedisVL, and similar solutions.
WebLLMs and locally-run LLMs at the edge will gain prominence. With the rise of Small Models and the ubiquity of AI functionality, applications will perform inference locally on devices, thanks to WebLLMs. This shift eliminates the need for remote server calls, enabling seamless, browser-based AI experiences without installations or configurations.
30 AI Libraries for the Modern AI Stack
In December, we were thrilled to introduce the first 30 open-source libraries as part of The Modern AI Stack (🌽MAIS). This collection of emerging tools aims to radically simplify building AI applications by abstracting away infrastructure complexity.
For example, orchestration frameworks handle intricate prompt engineering tasks like chaining multiple API calls, maintaining state across conversations, and retrieving relevant data from sources like vector databases. Monitoring tools provide end-to-end observability of model performance in production. Specialized vector databases efficiently store and retrieve contextual embeddings to augment model inputs.
Together, these libraries streamline, incorporating sophisticated AI capabilities like natural language processing, computer vision, recommendations, and search into apps built by professional developers and novice makers. As Andrej Karpathy notes, "Pre-trained AI models represent the most important architectural change in software since the internet." However, fully realizing their potential requires new infrastructure - that's what the 🌽MAIS aims to provide.
Over the coming year, we plan to highlight 30 new libraries every month across categories like pipelines, model operations, app building, and trust & security. For instance, this month's set includes:
LiteLLM (YC W23) - Simplified interface to all LLM APIs
Chroma - Painless local vector search and storage
LMQL - Improved control and safety for LLM integration
LLaMa-Factory - Rapid customized LLM development
Marvin by Prefect - Scalable production-grade NLP systems
Qdrant - Blazing fast vector similarity search
We are covering the full breadth of the emerging stack - from data connectors to model tooling to front-end interfaces. We choose libraries based on usability, modular architecture, performance, and impact potential.
We welcome suggestions on libraries you'd like to see featured. What AI development challenges do you face? Which new capabilities would be most valuable to accelerate your product roadmap? Let us know so we can expand 🌽MAIS to empower more developers.
The 🌽MAIS represents a vision for a blossoming ecosystem of interoperable building blocks that make AI radically more accessible. By sharing best-of-breed libraries, we want to enable faster experimentation, transparent capabilities, and impactful real-world solutions. Too often today, critical infrastructure knowledge lives in siloed big tech companies. We envision a different paradigm - driven by community-driven innovation.
We invite you to join us each month as we release new Modern AI Stack libraries for building the next generation of AI applications. We can transform lofty AI dreams into everyday reality by pooling collective insight.
🌽MAIS December Edition
Autogen
Axflow
Chroma
DSPy
FlowiseAI
Haystack
Jina
Langfuse
LiteLLM
LLaMa-Factory
llamafile
LLM-VM
LM Format Enforcer
LMQL
Ludwig
Marvin
Milvus
MindsDB
Ollama
OpenLLM
pgvector
Qdrant
RAGAS
rags
Redisvl
Superagent
SuperDuperDB
unsloth
Weviate
ZenML
Autogen
Autogen by Microsoft provides the perfect blend of power and simplicity for engineers to leverage multiple LLMs and tools for advanced yet reliable applications.
👷🏽Builders: Chi Wang, Qingyun Wu, Julia Kiseleva, Negar Arabzadeh
🤔 Why should The AI Engineer care about Autogen?
🚀 Productivity - Orchestrate multiple LLMs and tools to tackle complex workflows 10x faster than solo models.
🎯 Specialization - Assign roles matched to each asset's strengths for higher quality output.
🔁 Reusability - Codified workflows, data, and prompts simplify building new use cases.
Axflow
Axflow (YC S23) is a modular TypeScript framework for robust AI apps. It unlocks order-of-magnitude leverage for AI engineers through modular, scalable building blocks.
👷🏽Builders: Nicholas Charriere, Ben Reinhart
🤔 Why should The AI Engineer care about Axflow?
🧩 Modularity - Axflow is intentionally designed as modular components that can be adopted incrementally. This means you only take the parts you need, avoiding bundle bloat. It also enables easily swapping components, like models or data connectors, with no code change.
⚡️ Speed - Between modular and reusable components like prompts and data helpers and integrated eval for rapid iteration, Axflow accelerates building production-grade LLM apps 3-5x faster than coding from scratch.
🔌 Flexibility - Axflow uses interface-driven development to change underlying implementations without breaking contracts. This means you can connect new data sources, swap model backends, or exchange deploy environments with no code change.
Chroma
Chroma provides an essential and painless building block for engineers to integrate vector search into large language model applications. By abstracting complexity, Chroma amplifies leverage for creating the next generation of AI-powered software.
👷🏽Builders: Jeff Huber, Luke VanderHart, Anton Troynikov, Trayan Azarov
🤔 Why should The AI Engineer care about Chroma?
📡 Simplicity - Zero config vector search and storage lets engineers focus on apps, not infra.
🔌 Modularity - Plugs right into LangChain, LlamaIndex, and other libraries with consistent API.
⚡ Performance - Designed for low latency querying needed in real-time LLM loops.
DSPy
DSPy greatly accelerates the process of constructing dependable, scalable software using large language models and retrieval methods. Automating the refinement of prompts substantially increases the efficiency and influence of AI engineers working on conversational artificial intelligence platforms.
👷🏽Builders: Omar Khattab, Sri Vardhamanan A , Manish Shetty, Shangyin Tan
🤔 Why should The AI Engineer care about DSPy?
🔭 Abstraction - Focus on system architecture, not prompt tuning. Cleanly separate concerns.
♻️ Reusability - Modules and programs can be reused across models and tasks. No duplication.
💪 Maintainability - Changing data or models won't break prompts. Just recompile.
FlowiseAI
FlowiseAI (YC S23) accelerates innovation and collaboration around large language models by making it more straightforward to build customized solutions. Its visual approach also makes LLMs more accessible to domain experts beyond just AI engineers.
👷🏽Builders: ZhenJing Heng (Henry), Chung Yau Ong
🤔 Why should The AI Engineer care about FlowiseAI?
🤝 Accessibility - Low-code workflow visually empowers less technical users to build customized LLM solutions.
🧩 Modularity - Lego-like prebuilt blocks enable combining capabilities, models, data, etc, into reusable flows.
⚡️ Speed - Interactive flow builder accelerates constructing functioning LLM backends 3-5x faster.
Haystack
Haystack by deepset provides the essential, battle-tested building blocks for engineers to rapidly build tailored, scalable, and governable natural language processing systems. By simplifying access to best practices, Haystack enables developer leverage for innovation.
👷🏽Builders: Malte Pietsch, Milos Rusic, Mathis Lucka
🤔 Why should The AI Engineer care about Haystack?
🧩 Modularity - Flexible components and pipelines enable custom solutions tailored to specific needs.
⚡️ Productivity - 10x faster development cycles building robust NLP systems vs coding from scratch.
📡 Simplicity - Consistent API and abstractions minimize the need to handle model complexities.
Jina
Jina AI provides the perfect blend of simplicity and depth for engineers to build reliable and scalable AI applications rapidly. With maximized leverage, engineers amplify their innovation capability dramatically.
👷🏽Builders: Han Xiao, Joan Fontanals Martínez , Nan Wang
🤔 Why should The AI Engineer care about Jina?
⚡️ Productivity - Build and deploy models faster with a robust microservices framework.
🔌 Flexibility - Hotswap components like models and data sources with no code change.
📡 Simplicity - Abstraction from infrastructure lets you focus on model logic.
Langfuse
Langfuse (YC W23) is open-source observability and analytics for LLM apps.
👷🏽Builders: Marc Klingen, Max Deichmann, Clemens Rawert
🤔 Why should The AI Engineer care about Langfuse?
📈 Langfuse provides end-to-end observability into large language model (LLM) based applications, enabling engineers to debug issues and understand how changes impact metrics like quality, cost, and latency. Monitoring production systems matters.
📊 The ability to trace executions connecting prompts to responses, add custom scoring, and segment data by numerous parameters delivers granular analytics on LLM apps. Analytics drives optimization.
🔌 Integrations with popular frameworks like Langchain, OpenAI SDK, and LiteLLM combined with Langfuse's batteries-included capabilities simplify instrumenting model serving systems. Easy instrumentation means more monitoring.
LiteLLM
LiteLLM (YC W23) eliminates the undifferentiated heavy lifting when leveraging LLMs so engineers can focus innovation on end-user functionality and capabilities.
👷🏽Builders: Ishaan Jaffer, Krrish D.
🤔 Why should The AI Engineer care about LiteLLM?
⚡️ Speed - 10x faster development cycles integrating any LLM API into apps
🔌 Flexibility - Hotswap LLMs and providers with no code change via a unified interface
📡 Simplicity - One line of code removes the need to handle model complexities
LLaMa-Factory
LLaMA Factory combines simplicity and depth for rapidly building customized LLMs. Whether prototyping new ideas or deploying specialized models, it's an essential tool for any AI engineer working at the forefront of conversational AI.
👷🏽Builders: Yaowei Zheng, Yuchen
🤔 Why should The AI Engineer care about LLaMa-Factory?
🚀 Rapid prototyping - Quickly build and iterate on LLMs for proof of concepts without infrastructure overhead. Faster experimentation means faster innovation.
⚙️ Customization - Fine-tune models precisely as you need through exposed parameters like batch size, learning rate, hyperparameters, etc. Tailor to your use case.
🧑🏫 Education - Understand the inner workings of techniques like LoRA tuning by tweaking configs. Ideal for hands-on learning.
llamafile
llamafile by Mozilla represents an innovative approach to distributing and running LLMs, solving several practical problems for AI engineers around portability, accessibility, experimentation, and preservation of their work. Its simplicity and elegance are noteworthy.
👷🏽Builders: Justine Tunney, Stephen Hood, Ziad Ben Hadj-Alouane
🤔 Why should The AI Engineer care about llamafile?
🚚 Portability and accessibility - Llamafile allows large language models to be distributed and run as a single executable file that works across multiple operating systems and hardware architectures. This makes deploying LLMs much more portable and accessible without installing dependencies or setting up environments.
🔒 Offline capabilities - Since llamafile packages everything into a self-contained binary, the models can run fully offline without needing a network connection or external services. This is useful for reliability and privacy reasons.
🧪 Prototyping and experimentation - Llamafile allows AI engineers to prototype and experiment with different LLMs locally by downloading and running a file. This enables faster iteration.
LLM-VM
LLM-VM by Anarchy (YC W23) handles the undifferentiated heavy lifting so engineers can rapidly build and iterate language-based products. It saves time and cost while providing guardrails and best practices for success with LLMs.
👷🏽Builders: Abhigya Sodani, Matthew Mirman, PhD, Carter Schonwald
🤔 Why should The AI Engineer care about LLM-VM?
🛠 Simplicity - Abstracts away infrastructure so engineers can focus on product logic and capabilities using LLMs versus managing complexity.
📦 Modularity - Swap out models, data sources, and APIs with no code changes. Great for testing ideas.
⚡️ Optimization - State-of-the-art batching, quantization, etc., which would be costly to build custom means better performance.
LM Format Enforcer
LM Format Enforcer provides guardrails for engineers to build consistent and robust solutions needing structured outputs from language models. By guaranteeing format compliance, quality and governance increase dramatically.
👷🏽Builders: Noam Gat, Benedikt Fuchs
🤔 Why should The AI Engineer care about LLM Format Enforcer?
📋 Output Control - Enforces precise text formats like JSON and regex from LLMs.
🔬 Observability - Logs token scores reveal model struggles from constraints.
🎯 Accuracy - Format compliance reduces costly invalid outputs.
LMQL
LMQL gives AI engineers more control over LLM prompting through constraints, decoding techniques, performance improvements, and backend portability. This makes integration safer while expanding what's possible.
👷🏽Builders: Luca Beurer-Kellner, Marc Fischer, Matthew Dupree
🤔 Why should The AI Engineer care about LMQL?
🧱 Modularity - Break large prompts into reusable, configurable components. Build libraries of prompt modules.
⛓️ Constraints - Better control over LLM responses with token lengths, data types, regexes, etc. Critical for safety.
🔎 Decoding - Sophisticated decoding algorithms like beam search to deeply explore reasoning chains.
Ludwig
Ludwig by Predibase is an open-source low-code framework for easily building custom AI models like LLMs. It brings productivity to engineers through radically simplified access to best practices in deep learning.
👷🏽Builders: Piero Molino, Travis Addair, Devvret Rishi , Justin Zhao
🤔 Why should The AI Engineer care about Ludwig?
⚡️ Productivity - Faster development cycles building models with no boilerplate code.
📊 Governance - Standardized benchmarks and metrics aid model quality enforcement.
🧩 Modularity - Flexible components enable custom solutions tailored to specific needs.
Marvin
Marvin by Prefect enables AI engineers to build scalable, production-grade NLP systems by abstracting away almost all the undifferentiated heavy lifting. This multiplied leverage allows one to focus innovation purely on user-facing capabilities.
👷🏽Builders: Jeremiah Lowin, Nate Nowack, Adam Azzam, PhD, Jeff Hale
🤔 Why should The AI Engineer care about Marvin?
💻Developer Experience - LLMs feel like regular Python—no context switching from prompts.
📦 Modularity - Components are reusable across models with no code changes needed.
💪 Reliability - Handles errors, monitoring, and caching, unlike manual prompting.
Milvus
Milvus by Zilliz is an open-source cloud-native vector database built for high-performance embedding similarity search and AI apps.
👷🏽Builders: Charles Xie, Frank Liu, Yujian Tang
🤔 Why should The AI Engineer care about Milvus?
👉 It provides a millisecond search on trillion vector datasets for fast retrieval and recommendations. This allows AI engineers to build scalable applications with real-time responses. 🚀
👉 It offers simplified unstructured data management with rich APIs tailored for data science workflows. This makes it easier for AI engineers to work with embedding data at scale. 🧑💻
👉 It delivers consistent user experience across devices, allowing AI engineers to embed real-time search in any application. Whether working locally or in the cloud, the experience stays the same. 💻
MindsDB
MindsDB is an AI automation platform connecting data sources to LLMs and ML models for automated predictions.
👷🏽Builders: Jorge Torres, Adam Carrigan
🤔 Why should The AI Engineer care about MindsDB?
⚡️MindsDB dramatically simplifies building end-to-end AI applications by providing a unified SQL interface to connect data sources to LLMs and classic ML models. Abstraction enables faster development.
🚀 The extensive library of 130+ native integrations into data stores and the ability to auto-generate tailored models accelerate prototyping ideas. Quick prototyping drives creativity.
🤖 Event-based automation features streamline taking prototypes to production-grade systems with minimal additional work by developers. Automation ensures reliability at scale.
Ollama
Ollama enables AI engineers to build rapidly and iterate language model-based applications without cloud vendor lock-in. Making local LLM deployment push-button simple across platforms unlocks creativity and innovation.
👷🏽Builders: Jeffrey Morgan, Michael Chiang, Bruce MacDonald, Matt Williams, Patrick Devine
🤔 Why should The AI Engineer care about Ollama?
🤓 Abstract complexity - Handles infrastructure so engineers focus on product capabilities, not ops.
🔒 Privacy - Run models locally instead of sending data to third parties.
💰 Cost - Avoid paying for usage and egress bandwidth to cloud services.
OpenLLM
OpenLLM by BentoML provides all the infrastructure to go from model to production quickly while retaining the flexibility to customize and iterate. This allows engineers to focus innovation on the user experience with 10x leverage over building in-house.
👷🏽Builders: Chaoyu Yang, Anh Pham
🤔 Why should The AI Engineer care about OpenLLM?
🚀 Abstraction - Handles serving, scaling, and monitoring so engineers focus on building capabilities instead of infrastructure.
🧩 Modularity - Swap models, backends, hardware, and integrate tools like LangChain with no code change.
⚡️ Performance - State-of-the-art optimizations like streaming, batching, and quantization customized per model.
pgvector
pgvector simplifies the data side for AI engineers by tightly coupling vector capabilities into the underlying data management platform. This allows engineers to focus innovation higher up the stack on models, algorithms, and applications.
👷🏽Builders: Andrew Kane, Jonathan Katz
🤔 Why should The AI Engineer care about pgvector?
🤝 Tight integration with data infrastructure - Storing vectors directly in Postgres eliminates extra movement/coordination between separate vector search and data storage systems. Engineers can focus more on models and less on data engineering.
📊 Leverage relational power - Ability to quickly join vectors to other tables for added context, filtering, analysis, etc. SQL is also great for vector aggregations/analytics.
📈 Scale vectors massively - With partitioning and Postgres' proven scalability, it can store tens of billions of vectors queried in real-time.
Qdrant
Qudrant is a blazing-fast vector database and search engine optimized for embedding similarity queries. Filter results on rich payload data types. Scales distributed. It integrates with LangChain, LlamaIndex, OpenAI, and more, all coded in Rust.
👷🏽Builders: Andrey Vasnetsov, Tim Visée, Ivan Pleshkov, Egor Ivkov, Roman Titov, Kacper Łukawski
🤔 Why should The AI Engineer care about Qdrant?
👨💻 It provides an easy-to-use API to build neural search applications quickly. Less time fussing with infrastructure means faster development.
🚀 It offers state-of-the-art speed for vector similarity search to power real-time recommendations and matching. Fast response times keep users engaged.
⚡️ It scales horizontally to handle growing data volumes cost-efficiently—no need to overprovision hardware upfront. Add nodes as needed.
RAGAS
Ragas empowers engineers to build reliable and transparent RAG-based AI systems by providing a comprehensive toolkit for evaluation and monitoring.
👷🏽Builders: Jithin James, Shahul ES, Tino Max Thayil, Hwang yongtae, Armando Diaz
🤔 Why should The AI Engineer care about Ragas?
📊 Ragas makes evaluating and monitoring retrieval augmented generation (RAG) systems built using large language models (LLMs) dramatically more robust and reproducible. Rigorous evaluation methodology matters as we build more powerful assistants.
🔬 Capabilities like generating multi-faceted synthetic test data and quantifying metrics on aspects like hallucination enable engineers to diagnose weaknesses and incrementally strengthen systems. Targeted incremental improvement drives progress.
⚙️ Integrations with MLOps platforms such as LangFuse streamline instrumenting Ragas metrics as part of continuous integration, allowing rapid detection of regressions. Automated regression testing prevents nasty surprises.
RAGS
RAGS by LlamaIndex artfully blends ease of use with advanced capabilities for swiftly building bespoke RAG systems. It streamlines mocking up proofs of concept and deploying highly customized solutions, making it an indispensable asset for AI engineers operating on the frontier of natural language processing. Whether iterating new ideas or actualizing specialized designs, RAGS empowers progress at the forefront.
👷🏽Builders: Jerry Liu, Logan Markewich, Alexandros Filothodoros
🤔 Why should The AI Engineer care about RAGS?
⚡️ Fast testing: RAGS allows swiftly trying out RAG concepts and proofs of concept without needing to code an entire pipeline. The straightforward configuration through natural language facilitates exploration.
⚙️ Adjustability: AI developers can really customize RAG parameters such as lookup, summarization, embedding models, LLMs, etc. This adaptability gives fine-grained control to construct the optimal architecture.
📈 Scalability: Since RAGS handles the underlying pipeline building blocks, engineers can concentrate on higher-level challenges like scaling, deployment, and integration. Constructing custom frontends is also simpler.
Redisvl
RedisVL by Redis provides a Python client library to leverage Redis as a vector database.
👷🏽Builders: Sam Partee, Tyler Hutcherson, Chayim Kirshen
🤔 Why should The AI Engineer care about Redisvl?
👩💻 It provides an easy-to-use Python client library to build vector similarity search applications quickly. Less time fussing with infrastructure means faster development. 🚀
⚡️ It leverages the speed and reliability of Redis to deliver high-performance vector search to power real-time recommendations and matching. Sub-millisecond response times keep users engaged.
🌈 It simplifies working with embeddings by integrating directly with top vectorization services like OpenAI, HuggingFace, and VertexAI. Create vectors without needing to run complex ML training jobs.
Superagent
Superagent lets engineers build agents quickly with reduced complexity. By handling the undifferentiated heavy lifting, Superagent allows one to focus innovation on the user experience and problem-solving.
👷🏽Builders: Ismail Pelaseyed, Alan Zabihi
🤔 Why should The AI Engineer care about Superagent?
⚡️ Speed - Go from idea to production AI assistant in minutes instead of months.
🔭 Abstraction - Focus on capabilities rather than infra or ops.
⚙️ Flexibility - Easily swap out models, data sources, and tools. Great for testing ideas.
SuperDuperDB
SuperDuperDB empowers engineers to take ideas swiftly from conception to operationalization in real-world systems by simplifying the process of enhancing databases with AI capabilities.
👷🏽Builders: Timo Hagenow, Duncan Blythe, Tom Ritchford, Fotis Nikolaidis
🤔 Why should The AI Engineer care about SuperDuperDB?
⚡️SuperduperDB dramatically accelerates building end-to-end AI apps by integrating models directly into databases - no migration or duplication needed. Reduced overhead enables faster experimentation.
👩🔬 Streamlining capabilities like in-DB training, streaming inference, and vector search using simple Python minimizes the specialized skills needed. Democratization drives broader innovation.
🧩 Chaining multiple models into Python workflows without added infrastructure complexity enhances extensibility and transparency—modular architecture matters.
unsloth
Unsloth by Moonshot gives AI engineers the two most precious resources - time and money. Whether researching cutting-edge techniques or deploying LLMs at scale, faster innovation and reduced costs will drive impact.
👷🏽Builders: Daniel Han
🤔 Why should The AI Engineer care about unsloth?
⚡️ Rapid Prototyping - Quickly iterate and test ideas with up to 30x faster fine-tuning turnaround times.
💰 Cost Savings - Train models faster and with less GPU memory for dramatically lower compute costs.
🏭 Productization - Speed up research to production rollout of LLM models and capabilities.
Weaviate
Weaviate is a fast, scalable, cloud-native open-source vector database for building semantic search apps powered by state-of-the-art AI models.
👷🏽Builders: Etienne Dilocker, Bob van Luijt, Dirk Kulawiak, Marcin Antas, Connor Shorten
🤔 Why should The AI Engineer care about Weaviate?
👩💻 It provides an easy-to-use platform to build AI applications quickly without creating boilerplate code or managing infrastructure. Less time on setup means faster development. 🚀
⚡️ It offers lightning-fast vector similarity search to power real-time recommendations and matching. Sub-millisecond response times keep users engaged.
🧩 It enables hybrid search combining vectors, filters, and full-text search for flexibility to tailor the search experience. More options to build sophisticated use cases
ZenML
ZenML is an extensible, open-source MLOps framework for creating portable, production-ready machine learning pipelines.
👷🏽Builders: Hamza Tahir, Adam Probst , Alex S.
🤔 Why should The AI Engineer care about ZenML?
⚡️ ZenML simplifies building end-to-end MLOps pipelines from experimentation to production deployment. As AI practitioners, efficient workflow and infrastructure accelerate our development and impact.
🤝 The open-source nature of ZenML, combined with its integration across 50+ MLOps tools, provides flexibility in leveraging existing and emerging solutions as part of a cohesive workflow. This allows for staying on the cutting edge.
👥 ZenML's architecture that decouples code from infrastructure enhances collaboration across data scientists, ML engineers, and ops specialists. Cross-functional coordination is critical for impactful model development.