Bloomy (YC S26) 正在招聘创始工程师

1 分•作者: alexsouthmayd•3 个月前
关于 Bloomy Bloomy 是一个面向 K-12 学生的人工智能导师和课程。在早期试点中,我们的学生用户学习速度大约是原来的两倍,学生和老师都非常喜欢我们的产品。 工作原理:我们诊断学生的技能差距,为他们规划个性化的学习路径,然后由人工智能导师进行教学。AI 会学习他们的学习方式,从而提高教学效率,进而提高学习效率。 您如何提供帮助:您将作为创始工程师加入。目前,工程团队只有一位创始人(我),他通过协调大量的编码代理来根据规范、验证工具和质量保证门来生成大部分代码和课程。我是一位拥有前教师和销售领导者领域专业知识的独立创始人,但我并非科班出身的工程师,我需要一位真正具备技术专长的人来实现改善数百万学生教育的愿景。 我们为何与众不同:Bloomy 是原生 AI。我从今年 2 月到 4 月构建了整个产品,从 4 月到 6 月将其提供给 6 所学校的 500 名学生使用,6 月份开始参加 YC,此后 ARR 大约每两周翻一番。今年秋季,将有数千名学生使用我们的产品。Bloomy 也是过去五个批次中第一家通过 YC 的 EdTech 公司,因此这是一个机会,可以成为少数几家试图用 AI 改变教育的初创公司之一。 您最初的 90 天 第一周 将一个功能或修复程序发布到生产环境,并在第二天观察真实学生在真实课堂中使用它。我们从“合并”到“一名六年级学生使用它”的周期尽可能短。 第一个月 熟悉 React/TypeScript 应用、Supabase 后端、内容管道以及为其提供支持的代理集群。负责您的第一个集群——编写规范、构建工具、设置质量保证门——并通过它开始发布。 三个月后 与我一起设计路线图:约 50% 的时间用于跨堆栈发布产品,约 25% 的时间用于构建和运行代理基础设施,约 25% 的时间用于与教师、学生和学区交流——并将您听到的内容转化为路线图。 我们非常想构建的东西 - 数据飞轮。教育领域几乎没有人能够将遥测数据与教学本身联系起来。我们可以在几天内完成——针对每一次回答、重试、AI 导师对话以及任何其他互动——而不是经过教科书修订周期。设计这个循环。 - 将代理编排作为一项一流的工程问题。更好的工具、更好的评估、更好的验证——这样,十几个并发代理就能生成您信任的、可以在课堂上使用的工作。 - 学校接下来需要的东西。今天我们为学生和教师提供工具。学区的表面积巨大,您将帮助决定我们的发展方向。 我们正在寻找这样的人: - 拥有 5 年以上交付真实用户日常使用的产品经验,并对如何使产品优秀有自己的见解。 - 主动性强:能够接受模糊的反馈并将其转化为已发布的产品,而无需等待规范。 - 精通 AI 编码工具,并利用它们来显著提高效率——而不是将判断外包给它们。 - 对管理十几个并发代理的想法感到兴奋,而不是精疲力尽。 - 希望与真实用户(其中一些是十一岁的孩子)建立紧密的反馈循环。 - 具有很强的抗压能力——花名册 CSV 导入、学校设备性能怪癖以及学校特定的定制都是工作的一部分,也是我们获胜的原因。 - 关心可衡量的学习成果,而不仅仅是参与度指标。 优先经验 - TypeScript、React、Postgres(我们使用的是 React 18、Vite、Tailwind、shadcn/ui、Supabase、TanStack Query,部署在 Vercel 上)。 - 生产环境中的 LLM:代理、评估、工具使用、上下文管理。 - 构建供代理使用的代码库和管道。 - 事件遥测、分析或实验系统。 - Edtech 背景并非必需。 背景故事 我的职业生涯始于一名七年级英语老师,我亲眼目睹了教授 30 个不同理解水平的 30 个孩子的艰辛。在 Lyft 构建了司机获取引擎,在斯坦福攻读 MBA 期间进行了各种创业尝试,并在麦肯锡从 AI 代理和工作流程的兴起之初就开始构建它们,我现在相信教育是 AI 可以对社会产生最大积极影响的领域。 薪酬 - 175,000 美元 - 200,000 美元 - 1.5% - 2.0% 的股权 创始工程师套餐——从第一天起就拥有有意义的所有权。 听起来像你吗? 发送电子邮件至 alex@bloomylearning.com。请附上您构建过的、真实用户使用过的东西。
查看原文
<i>About Bloomy</i><p>Bloomy is an AI tutor and curriculum for K-12 students. Our student users are learning approximately 2x faster in early pilots, and students and teachers alike love our product.<p>How it works: We diagnose students’ skill gaps, map out a personalized learning path for them, and then have an AI tutor teach them. The AI learns about how they learn, so it can increase the efficiency of its teaching and, consequently, their learning.<p>How you can help: You will be joining as the founding engineer. Right now, engineering is one founder (me) orchestrating fleets of coding agents that produce a large share of our code and curriculum against specs, verification harnesses, and QA gates. I’m a solo founder with domain expertise as a former teacher and sales leader, but I’m not an engineer by training, and I need someone with true technical expertise to actualize the vision of improving education for millions of students.<p>Why we’re different: Bloomy is AI-native. I built the entire product from February to April of this year, got it into the hands of 500 students across 6 schools from April to June, started YC in June, and have roughly doubled ARR every 2 weeks since then. Thousands of students will soon use our product this fall. Bloomy is also the first EdTech company to go through YC for the past 5 batches, so it represents an opportunity to be on the ground floor of one of relatively few startups trying to transform education with AI.<p><i>Your first 90 days</i><p><i>First week</i><p>Ship a feature or a fix to production and watch real students in real classrooms hit it the next school day. Our loop from “merged” to “a 6th grader used it” is as short as it gets.<p><i>First month</i><p>Get acquainted with the React&#x2F;TypeScript app, the Supabase backend, the content pipeline, and the agent fleets that feed it. Take ownership of your first fleet — write the spec, build the harness, set the QA gate — and start shipping through it.<p><i>Three months later</i><p>Design the roadmap with me: ~50% shipping product across the stack, ~25% building and running agent infrastructure, ~25% talking to teachers, students, and districts — and turning what you hear into the roadmap.<p><i>Things we badly want to build</i><p>- The data flywheel. Almost nobody in education closes the loop between that telemetry and the instruction itself. We can — for every answer, retry, AI-tutor conversation, and any other interaction — in days, not textbook-revision cycles. Design that loop.<p>- Agent orchestration as a first-class engineering problem. Better harnesses, better evals, better verification — so that a dozen concurrent agents produce work you&#x27;d trust in front of a classroom.<p>- Whatever schools need next. Today it&#x27;s tools for students and teachers. The surface area in a school district is enormous, and you&#x27;ll help decide where we go.<p><i>We&#x27;re looking for someone who</i><p>- Has 5+ years shipping products that real end users touch daily, and has opinions about what makes them good<p>- Is high-agency: can take ambiguous feedback and turn it into shipped product without waiting for a spec<p>- Is expert-level with AI coding tools and uses them to move dramatically faster — without outsourcing judgment to them<p>- Gets excited, not exhausted, by the idea of managing a dozen concurrent agents<p>- Wants tight feedback loops with real users (some of whom are eleven)<p>- Has high schlep tolerance — roster CSV imports, school-device performance quirks, and school-specific customization are part of the job and part of why we win<p>- Cares about measurable learning outcomes, not just engagement metrics<p><i>Preferred experience</i><p>- TypeScript, React, Postgres (we&#x27;re React 18, Vite, Tailwind, shadcn&#x2F;ui, Supabase, TanStack Query, on Vercel)<p>- LLMs in production: agents, evals, tool use, context management<p>- Building codebases and pipelines designed to be worked on by agents<p>- Event telemetry, analytics, or experimentation systems<p>- Edtech background genuinely not required<p><i>The backstory</i><p>I started my career as a 7th grade English teacher, where I saw how hard teaching 30 different kids at 30 different levels of understanding really is. After building the driver acquisition engine at Lyft, pursuing various entrepreneurial ventures while getting my MBA at Stanford, and building AI agents and workflows at McKinsey from the start of the boom, I now believe education is where AI can have the largest positive impact for society.<p><i>Compensation</i><p>- $175K - $200K<p>- 1.5 - 2.0% equity<p>Founding-engineer package — meaningful ownership from day one.<p><i>Sound like you?</i><p>Email alex@bloomylearning.com. Include something you&#x27;ve built that real people used.