Agentic Engineering Intern at edon.ai


Remote first, with a weekly workshop in Kaunas.

edon.ai translates and localises video and other training artefacts. That sounds like a solved problem. It isn’t. Generic machine translation will happily render the same term three different ways in one video. Our customers are media and corporate training companies, and they can’t live with that.

So we give them control. Glossaries, idioms and phrasing hold in every language, every time.

Now we’re building the video side. Customers will be able to describe a new scene, have AI generate it and cut it into a video they already have.

We’re looking for an intern to work on both.

The work

Most of it is the translation engine. Glossaries and idioms have to survive every language we support. You’ll build agents that catch slips before a human reviewer has to. You’ll also write the evals that tell us whether a change made a translation better or just different. That second part matters more than it sounds.

Some of it is scene generation. A user describes a scene, a model makes it, and it has to sit in their video without looking pasted in. That’s harder than generating a nice clip.

This is real work on a product people pay for. It isn’t a side project we made up to keep an intern busy.

How we work

We’re a small team and we work remotely. Most of our thinking happens in writing, so how you write matters as much as how you code.

We don’t do hand-holding. You get a problem, access to the code and people who answer questions quickly. Then it’s yours. If that sounds stressful, this probably isn’t for you. If it sounds like a relief, keep reading.

Once a week we meet in person in Kaunas. You show what you built, we poke holes in it, and we decide what’s next.

Who we’re looking for

You already write code with Claude Code or Codex. You know what shipped this month and you have an opinion on whether it was any good. You’ve built something with LLMs because you were curious, not because a module required it.

You’re a generalist. You might do backend one day, a bit of frontend the next, and a script, a prompt or a spreadsheet when that’s what the problem calls for. You’re comfortable in Python or TypeScript and willing to learn the other.

You ask good questions, then go and find the answers. You don’t wait to be told what’s next.

We don’t care much about grades. We care about what you’ve made.

Who we’re not looking for

We’re not a match if you need a full spec before you start, or if “that’s not my job” comes naturally to you. The same goes if you insist on writing every line by hand, or think agents are just chatbots with a marketing budget. No hard feelings.

After the internship

If it works for both of us, we’ll offer you a job. We’re not running an internship for the sake of it. We’re looking for someone to keep.

How to apply

Send us a link to something you built with AI: a repo, a demo, a video, anything that runs. Then write three sentences about what broke and how you fixed it.
Email: labas@edon.lt

Those three sentences are your first writing sample. We’ll read them closely.

Norėdami sužinoti daugiau apie siūlomą praktikos vietą, galite susisiekti su įmonės kontaktiniu asmeniu:

Atsakingas asmuo Samanta Bulovaitė (Studentų patirties koordinatorė)
Tel. Nr. (+370...) +37066811011
El. paštas samanta.bulovaite@edon.lt