AI Software Engineer Intern jobs in Kepler - Maukerja

About Kepler

Kepler (*************) is an AI-native CRM for recruitment agencies. Recruiters use it to manage candidates, clients and jobs. AI does the slow work: it reads CVs, fills in records, matches candidates to jobs, searches years of notes and emails, and drafts outreach. Agencies run their whole day on it.

About the internship

This is an AI engineering internship. You won't be working in a sandbox. You'll build AI features that recruiters use every day, on real data, where mistakes cost a placement. You'll find out what it takes to make an LLM feature work reliably, and that's hard to learn anywhere else as a student.

What you'll work on

• Evaluations. Build and run evals that compare models on real tasks like candidate matching and CV parsing, scoring them on quality, speed and cost. Your results will decide which model goes live.
• Retrieval and search. Work on embeddings, vector search and hybrid search across millions of CVs, notes and emails, where each agency's data is kept separate.
• Structured extraction. Turn messy CVs, PDFs and old CRM exports into clean, reliable records using structured output and validation.
• Agents and tools. Build assistants and integrations that take real actions inside the CRM, with guardrails so they act safely.
• Cost and scale. Design AI pipelines that stay cheap and fast for hundreds of agencies, not just one demo.
• Full stack. Ship each feature end to end, from React UI to Node.js API to Postgres, and prove it works with end-to-end tests.

What you'll gain

• Production AI experience. You'll ship to real users within your first weeks and see how they use your work.
• Judgement about models. You'll learn when to use a small model or a large one, when batch processing beats real-time, and when retrieval beats a longer prompt, all measured on real data.
• Ownership. You'll own problems from start to finish. Small team, direct work with the founder, no layers in between.
• An AI-native workflow. We build with AI coding agents every day, and you'll learn to use them well on a real codebase.
• Something to show. You'll leave with features in production you can talk about in any interview.

What we're looking for

• Currently pursuing or recently completed a relevant degree, ideally from leading universities such as NUS, NTU, SUTD, and SMU. Strong candidates from other institutions will also be considered.
• Solid JavaScript or TypeScript. Python is a plus
• You've built something with an LLM API: a side project, a hackathon entry, a tool for yourself
• You understand the basics of embeddings, retrieval or prompt design, or you're eager to learn them fast
• You test your own work and care whether it actually works, not just whether it demos well
• You can work with incomplete information and you'll ask when you're stuck

Nice to have

• Experience with PostgreSQL, React or Node.js
• You've built an eval or benchmark, even a rough one
• Interest in recruitment, sales tools or B2B SaaS

Details

• Duration: 3-6 months
• Start: Earliest convenience
• Location: Remote

To apply

Attach your resume and a link to something you've built with AI. In two sentences, tell us what broke while you were building it and how you fixed it.