About the Challenge
The BatchIn Build Challenge is a week-long, fully asynchronous competition hosted on Devpost. Developers around the world can participate on their own schedule — no live attendance required. Your mission: build the most impressive AI agent using BatchIn's inference APIs.
This isn't about speed. It's about craft. We want to see thoughtful agent architectures, creative use cases, and production-ready code.
关于本次挑战赛
BatchIn 构建挑战赛是一场为期一周、完全异步的编程比赛，托管在 Devpost 上。全球开发者可以按照自己的节奏参与——无需实时在线。你的任务：使用 BatchIn 推理 API 构建最令人印象深刻的 AI Agent。
这不是比速度。我们要看到深思熟虑的 Agent 架构、创新的应用场景和可用于生产的代码。
Challenge Rules
- Team Size: Solo or teams up to 3 people.
- Timeline: July 1 (00:00 UTC) — July 7 (23:59 UTC).
- Submission: Via Devpost. Must include a demo video (max 3 min), source code (GitHub), and a brief writeup.
- Requirements: Must use at least one BatchIn inference API (chat, batch, or embedding). Can combine multiple models.
- Judging Criteria: Innovation (30%), Technical Execution (30%), Usefulness (20%), Presentation (20%).
比赛规则
- 团队规模：个人或最多 3 人组队。
- 时间线：7 月 1 日 (00:00 UTC) — 7 月 7 日 (23:59 UTC)。
- 提交方式：通过 Devpost。需包含演示视频（最长 3 分钟）、源代码（GitHub）和简要说明。
- 要求：必须使用至少一个 BatchIn 推理 API（聊天、批量或向量嵌入）。可以组合多个模型。
- 评判标准：创新性 (30%)、技术实现 (30%)、实用性 (20%)、演示质量 (20%)。
Prizes
| Place | Prize |
|---|---|
| 🥇 1st Place | $400 API Credits + Featured on BatchIn Blog |
| 🥈 2nd Place | $250 API Credits |
| 🥉 3rd Place | $150 API Credits |
| 🏅 Honorable Mentions (2) | $100 API Credits each |
Total Prize Pool: $1,000 in API Credits
奖池
| 名次 | 奖品 |
|---|---|
| 🥇 第一名 | $400 API Credits + BatchIn 博客专题报道 |
| 🥈 第二名 | $250 API Credits |
| 🥉 第三名 | $150 API Credits |
| 🏅 荣誉提名 (2名) | 每人 $100 API Credits |
总奖池：$1,000 API Credits
Example Project Ideas
- Research Agent: An agent that takes a topic, searches academic papers, synthesizes findings, and generates a structured report.
- Code Review Agent: An agent that reviews pull requests, identifies bugs, suggests improvements, and auto-generates test cases.
- Customer Support Agent: A multi-turn agent that handles support tickets by reasoning through knowledge base articles and escalating when needed.
- Data Pipeline Agent: An agent that ingests CSV/JSON data, cleans it, runs analysis via batch API, and generates visual reports.
项目创意示例
- 研究 Agent：输入一个主题，搜索学术论文，综合发现，生成结构化报告。
- 代码审查 Agent：审查 Pull Request，识别 Bug，建议改进，自动生成测试用例。
- 客服 Agent：多轮对话 Agent，通过推理知识库文章处理工单，必要时升级处理。
- 数据流水线 Agent：接收 CSV/JSON 数据，清洗、通过批量 API 分析，生成可视化报告。
Judges
- Andy Wang — Co-founder & CEO, BatchIn.
- Lisa Park — Developer Advocate, BatchIn. Previously at Vercel and Hugging Face.
- Community Judge — Top voted participant from a previous BatchIn event.
评委
- Andy Wang — BatchIn 联合创始人 & CEO。
- Lisa Park — BatchIn 开发者布道师。曾任职于 Vercel 和 Hugging Face。
- 社区评委 — 往期 BatchIn 活动中最受欢迎的参与者。
FAQ
Q: Do I need to be online at specific times? A: No. This is fully asynchronous. Work whenever and wherever you want during the 7-day window.
Q: Can I use other APIs alongside BatchIn? A: Yes! You can use any tools, libraries, or external APIs. The only requirement is that your agent must use at least one BatchIn API.
Q: Can I submit a project I already started? A: Your project must be started after the challenge begins. Existing libraries and boilerplate are fine, but the agent logic must be new.
Q: How do I get started? A: Sign up for a free BatchIn account, grab your API key, and check out our quickstart guide at luminapath.tech/docs.
常见问题
问：我需要在特定时间在线吗？ 答：不需要。这是完全异步的。在 7 天窗口期内随时随地工作。
问：我可以同时使用其他 API 吗？ 答：可以！你可以使用任何工具、库或外部 API。唯一要求是你的 Agent 必须使用至少一个 BatchIn API。
问：我可以提交之前已经开始的项目吗？ 答：项目必须在挑战赛开始后启动。可以使用现有的库和脚手架，但 Agent 逻辑必须是新的。
问：如何开始？ 答：注册免费 BatchIn 账户，获取 API 密钥，查看我们在 luminapath.tech/docs 上的快速入门指南。