End-to-End App Development Made Easy with Devika - Your Personal AI powered Software Engineer

The open source alternative of Devin

Original post from my newsletter(May 18, 2024) — High Signal AI — subscribe to get this delivered to your inbox every week!

Action prompting insights:

1. Top find of the week 🏆: Devika — the open source coding agent
2. What people are building with AI!
3. GitHub repositories you must check out!
4. AI startups that raised funds
5. AI tutorials and videos on YouTube worth your time
6. Reading list for the week
7. Weekly updates from tools, libraries and hubs

Top find of the week 🏆: Devika

• Repo: stitionai/devika (17223 stars)
• Description: Devika is an Agentic AI Software Engineer that can understand high-level human instructions, break them down into steps, research relevant information, and write code to achieve the given objective. Devika aims to be a competitive open-source alternative to Devin by Cognition AI.
• Topics: Agents

Middleware Open-Source — An Engineering Productivity tool

• Middleware is an open-source tool designed to help engineering leaders measure and analyze the effectiveness of their teams using the DORA (DevOps Research and Assessment) metrics.
• It offers DORA metrics out of the box to start with — a non-monitoring metrics framework
• It is open source and free.
• Accessible for all types of orgs
• ✅ You can self-host it in your infra — Extremely secure
• ✅ Repo: https://github.com/middlewarehq/middleware

What people are building with AI!

• Using LangChainNatural Language-Assisted Visualization & Interactive Data Analysis (NLAVIDA): Securely handle and analyze confidential data in enterprise environments, enhancing insights and decision-making.
• Project link: https://github.com/obaidur-rahaman/nlavida
• A curated collection of awesome LLM apps built with RAG and AI agents. This repository features LLM apps that use models from OpenAI, Anthropic, Google, and even open-source models like LLaMA that you can run locally on your computer.

GitHub repositories you must check out!

• Repo: stitionai/devika (17223 stars)
• Description: Devika is an Agentic AI Software Engineer that can understand high-level human instructions, break them down into steps, research relevant information, and write code to achieve the given objective. Devika aims to be a competitive open-source alternative to Devin by Cognition AI.
• Topics: Agents
• Repo: decodingml/llm-twin-course (640 stars)
• Description: 🤖 𝗟𝗲𝗮𝗿𝗻 for 𝗳𝗿𝗲𝗲 how to 𝗯𝘂𝗶𝗹𝗱 an end-to-end 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻-𝗿𝗲𝗮𝗱𝘆 𝗟𝗟𝗠 & 𝗥𝗔𝗚 𝘀𝘆𝘀𝘁𝗲𝗺 using 𝗟𝗟𝗠𝗢𝗽𝘀 best practices: ~ 𝘴𝘰𝘶𝘳𝘤𝘦 𝘤𝘰𝘥𝘦 + 11 𝘩𝘢𝘯𝘥𝘴-𝘰𝘯 𝘭𝘦𝘴𝘴𝘰𝘯𝘴
• Topics: aws, bytewax, comet-ml, generative-ai, large-language-models, machine-learning-engineering, ml-system-design, mlops, qdrant, qwak, superlinked, course, docker, infrastructure-as-code, llmops, rag
• Repo: Beomi/InfiniTransformer — Implementation of the above mentioned InfiniAttention paper(249 stars)
• Description: Unofficial PyTorch/🤗Transformers(Gemma/Llama3) implementation of Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention
• Topics: gemma, huggingface, pytorch, transformers, infinitransformer, llama, llama3

AI startups that raised funds

The AI startups that recently received funding are:

1. Grasp (Stockholm) — Raised $1.9M for Investment Banking AI Assist.
2. Artisan AI (San Francisco) — Raised $7.3M for Enterprise AI.
3. Niqo Robotics (Bengaluru) — Raised $13M for AI-Powered Spot Spraying.

Source

AI tutorials and videos on YouTube to stay ahead of the curve!

• Umar Jamil made a great explanatory video on Kolmogorov-Arnold Networks, a new type of network that was presented in the paper “KAN: Kolmogorov-Arnold Networks”.
• AI Jason made a video about exploring universal web scraper agent that can use browser directly, e.g. pagination, catpcha, etc.
• A good conversation to get a quick overview of current state of RLHF and LLM evaluations.
• Want to know how microchips are made, here’s a video full of detailed visuals to help you understand the entire process.

Reading material that’ll help you get a deeper understanding of AI

caption…

• Well, definitely not something to be covered within a week, these were the 30 papers Ilya Sutskever supposedly sent to John Carmack to learn 90% of what matters today. You can build a solid AI research curriculum for yourself out of this.
• A book I came across for understanding deep learning, it has link to the GitHub repository filled with chapter-wise tutorials and examples for you to check out.

Weekly updates from tools, libraries and hubs.

• Cursor’s Llama3–70b Model: Cursor announced a new result, surpassing GPT-4 performance with speeds of over 1000 tokens/s using their speculative-decoding variant tailored for code edits.
• Open-Source Library — @llama_index added GPT-4o support in LlamaParse for complex document parsing and indexing.
• Paper in talks — InfiniAttention Paper: Discussed for its potential framework to improve LLM’s handling of long context without overwhelming finite working memory.
• Vertex AI Welcomes LlamaIndex: LlamaIndex teamed up with Vertex AI for a new RAG API, aiming to enhance users’ ability to implement retrieval-augmented generation models on Vertex’s cloud platform.

That’s a wrap! I hope you found these insights actionable. I’m dedicated to making this newsletter even more practical and engaging. Please feel free to suggest any topics or sections you’d like to see included in future editions.

Subscribe to High Signal AI — The most action oriented newsletter for developers and entrepreneurs building with AI!