数据录入实习生
Data Entry Intern
这是一个内部职位在周日。不是一个客户的角色。· 职位类型：实习 · 地点：远程 · 持续时间：3个月 ·工作日：星期一周六 · 奖学金：4,0005,000/月 · PPO 如果表现令人满意周日是人工智能原生招聘平台。我们由 Y 组合和风险高速公路 （现在通用催化剂）支持。我们建立了印度最大的白领人才数据库，并为他们运行出境招聘活动构建工具。在申请之前阅读这不是一种迷人的角色。这是谨慎的，高量度的，细节痴迷的工作，我们正在建造的模型只好像你生产的数据一样。你将花一天看文件，在正确的事情周围绘制边界框：标记一个字段，文本块，表或签名的位置，一次又一次地，正确。这是精确的重复工作，它是整个注释项目的基础。一个粗的盒子教导模型错误的事物。如果你从做某件事上得到满足感，完全正确，步伐，没有切断角落 你会在这里做得很好。如果你需要不断的多样性和厌倦重复，这不是你的。你实际上会做什么任何给定的日子你可能是：·查看文件，并绘制有关区域的精确界限框 字段，文字块，表格，邮票，签名，任何项目所需的东西 ·根据注释指南将每个框贴上标记和正确分类 ·保持你的框紧密一致 记录一项和文档的一千个标准 ·标志边缘案例，模糊的文件以及规则中的缺陷或矛盾而不是猜测 ·审查和纠正自己的 （有时其他）注释以保持高质量 ·帮助完善标签过程，以便在时间推移中更快地进行注释。你将以准确性，一致性和吞吐量测量你生产的高质量的注释数据，而不是记录时间。我们正在寻找什么 · 你是精确的和耐心的·你可以在第一个和第一个一次绘制相同的清洁盒子；详细内容很出汗；一个略有错误的盒子或错误的标签让你感到不安·你承担所有权。当一批货属于你的时，它是正确的·没有需要跟踪的 · 你紧随指南，但当一些东西不适合它们的时候说话 · 你对重复性的屏幕沉重的 Wo 感到舒服
查看英文原文
This is an internal role at Weekday. Not a role by a client.
- Role type: Internship
- Location: Remote
- Duration: 3 months
- Working days: Monday–Saturday
- Stipend: ₹4,000–5,000/month
- PPO if performance is satisfactory
Weekday is an AI-native recruiting platform. We are backed by Y Combinator and Venture Highway (now General Catalyst). We have built the largest white collar talent database in India and have built tools to run outbound recruiting campaigns to them.
Read this before you apply
This is not a glamorous role. It's careful, high-volume, detail-obsessed work — and the models we're building are only as good as the data you produce. You'll spend your day looking at documents and drawing bounding boxes around the right things: marking exactly where a field, a block of text, a table, or a signature sits on the page, over and over, correctly. It's precise, repetitive work, and it's the foundation the whole annotation project stands on. One sloppy box teaches the model the wrong thing.
If you get satisfaction from doing something exactly right, at pace, without cutting corners — you'll do well here. If you need constant variety and get bored by repetition, this isn't for you.
What you'll actually do
On any given day you might be:
- Looking at documents and drawing accurate bounding boxes around the regions that matter — fields, text blocks, tables, stamps, signatures, whatever the project calls for
- Labelling and classifying each box correctly according to the annotation guidelines
- Keeping your boxes tight and consistent — same standard on document one and document one thousand
- Flagging edge cases, ambiguous documents, and gaps or contradictions in the guidelines instead of guessing
- Reviewing and correcting your own (and sometimes others') annotations to keep quality high
- Helping refine the labelling process so annotation gets faster and more consistent over time
You'll be measured on accuracy, consistency, and throughput — how much high-quality annotated data you produce, not hours logged.
What we're looking for
- You're precise and patient — you can draw the same clean box the hundredth time as well as the first
- You sweat the details; a box that's slightly off or a wrong label bothers you
- You take ownership. When a batch is yours, it's done right — no follow-ups needed
- You follow guidelines closely but speak up the moment something doesn't fit them
- You're comfortable with repetitive, screen-heavy work and a Monday–Saturday pace
- You care about the quality of your work even when nobody's checking every box
What you get
- A front-row look at how a venture-backed startup is built and run profitably
- A real understanding of what goes into the data behind AI models — the part most people never see
- Real ownership of the ground truth the project depends on
- A PPO if you perform
- Hands-on experience with the tools and habits that make you genuinely useful anywhere
Originally posted on Himalayas