Jan 6
Memory Leak — #12
VC Astasia Myers’ perspectives on machine learning, cloud infrastructure, developer tools, open source, and security. Sign up here.
🚀 Products
HTTPie uses AI to helps individuals interact with APIs. You can describe a request using natural language (like “Fetch last release details of httpie/desktop”). Individuals can also ask the AI to generate an HTTP request by describing how the API or server works (like hostname, endpoints, auth) or low-level components (like the URL, headers, and body).
Why does this matter? We’ve discussed the rise of Large Language Models (LLMs), and its application to code generation like GitHub Copilot in the past. HTTPie AI uses OpenAI’s GPT-3 models and fits in the generative code category. It can be useful for non-technical users or software engineers. We expect there to be more AI software development solutions going forward. Both retrofitting existing products like HTTPie AI and new AI-first solutions built from the ground up with AI in mind.
SkyPilot is an open source framework for easily and cost effectively running ML workloads on any cloud. SkyPilot abstracts away cloud infra burden like 1) launch jobs & clusters on any cloud (AWS, Azure, GCP), 2) find scarce resources across zones/regions/clouds, and 3) queue jobs & use cloud object stores. SkyPilot supports existing GPU, TPU, and CPU workloads with no code changes.
Why does this matter? When speaking with ML practitioners the number one complaint you will hear is that ML model training and inference can be expensive. SkyPilot solves a key problem for teams.
Bun now offers better compatibility with Node APIs; bunx for auto-installs and running an executable from npm 100x faster than npx; improved stability; and enhancements to the Bun API.
Why does this matter? We are strong believers in the serverless edge movement. Bun has momentum in this space and continues to build compatibility with Node and Web APIs so developers more easily adopt it with their existing stack and tools.
📰 Content
What TNS Readers Want in 2023: More DevOps, API Coverage
The New Stack conducted a survey in November and early December 2022 on the topics readers are more interested and relevant to them in 2023. Across the 306 participants, API design/management/testing and DevOps are tied as the most popular topics, each cited by 55% of readers, followed closely by cloud infrastructure services (54%). APIs rose from number 10 in 2021 to number 1 in 2022. DevOps held onto the number one spot year over year. Interestingly platform engineering, like Humanitec, ranked for the first time at number 4 (52%).
Why does this matter? We are seeing a new wave of API development tooling from HTTPIe AI above to Speakeasy that generates client SDKs to achieve faster and easier integration for APIs. We’ve highlighted the emergence of platform engineering as a core category since summer 2022 so the appreciate additional validation of this trend.
WebAssembly: 5 Predictions for 2023
Matt Butcher, co-founder and CEO of Wasm cloud provider Fermyon, underscores that 2022 was a big year for Wasm since the Bytecode Alliance launched a number of Wasm standards. He states five 2023 predictions including: the component model will be the watershed moment; Wasm apps will be stored in DockerHub and container registries; and all the big programming languages will have WASM support.
Why does this matter? We are increasingly hearing about the use of WASM and think the creation of Wasm standards are a big step forward to improve adoption as well as compatibility between solutions. As Wasm sees more use cases across the edge, data center, and browser, particularly serverless use cases, we are excited by solutions that help users with day two operations like observing/monitoring, analyzing/troubleshooting, and maintaining/optimizing Wasm in production.
Lenny Rachitsky discusses how products actually go viral. He underscores that it’s not necessarily word of mouth, but “dark broadcasters’ — people or companies distributing information to many viewers at once, but whose influence isn’t always visible to people outside of the network.”
Why does this matter? This is a great piece for open source project maintainers and startup founders to read to better understand the most effective techniques to generate visibility and momentum around their work. The key takeaway: in the beginning, get high profile individuals like thought leaders or influencers within a community to advocate for your work.
💼 Jobs
⭐️Dragonfly — Developer Advocate (fully remote)
⭐️ Diagrid — Senior Backend Engineer
⭐️ Grit — Founding ML Engineer
