Careers

Come build AI that has to prove itself.

We are a small team in Ortigas building AI systems for enterprises that cannot send their data anywhere. That constraint makes the engineering harder and the work more interesting. Two roles are open, and two internship tracks for people still studying.

See open roles Internships

Why here

What you would actually get.

The work is measured: Every engagement agrees what good looks like before anything is built, and scores against it afterwards. You will know whether your work succeeded, which is rarer than it should be.; How that looks in practice →
Small team, real ownership: A project team is three or four people. There is no layer between you and the client’s problem, and no layer between your commit and production.; Who you would be working with →
We teach, so we have to understand it: PSI runs its own AI academy. Explaining a technique to a room of engineers is the fastest way to find out whether you actually know it.; PSI AI Academy →

Open roles

Two positions, both based in Pasig.

AI / LLM Engineer

Full-time · Mid-senior

You would be fine-tuning open-weight models to run inside a client’s own network, and proving they work well enough to be trusted there.

Ortigas, Pasig City. Hybrid, one day in the office each week.

What you would do

Fine-tune Llama, Gemma and DeepSeek models with PEFT and LoRA against a client’s own data.
Quantise and distil them down to something that runs on the hardware the client actually has.
Build retrieval over their documents with vector search, and measure whether it retrieves the right thing.
Ship the model behind an API that survives production traffic.
Write the evaluation that decides whether any of the above worked.

What we are looking for

Three or more years writing Python, with PyTorch or TensorFlow.
Hands-on fine-tuning experience, not just prompting: PEFT, LoRA, quantisation, distillation.
Hugging Face Transformers.
FastAPI, Celery and Redis, or the equivalent in another stack.
RAG and a vector database. We use Qdrant and FAISS.

Apply for this role

Front-end Developer

Full-time · Entry level

The interface is where an AI system either earns a user’s trust or loses it. This role builds that half.

Ortigas, Pasig City.

What you would do

Build product interfaces in React and Next.js.
Work against a Prisma-backed API, alongside the engineers who write it.
Turn designs into components that hold up on a real device, not just in the mockup.
Work in sprints, with the tester and the designer in the same room.

What we are looking for

Two or more years in front-end development.
JavaScript, HTML5 and CSS3, properly rather than through a framework you cannot see past.
React. Next.js and Prisma are what we use; we do not expect you to arrive knowing both.
Comfortable working in Scrum or another agile process.

Apply for this role

Internships

Learn before you graduate.

Most of what is worth knowing about building AI systems is not taught anywhere yet. You can wait until you graduate to find that out, or you can spend a few months inside a team that does it for a living.

We put interns on real client work rather than side projects invented to keep them busy. Two tracks, both open to students and to recent graduates.

UX testing Find out where real people get stuck, before a client does.: You would sit with the people who use what we build, watch where they hesitate, and write down what actually happened rather than what was supposed to. On live products, with real users.; Run usability sessions and take the notes that survive the meeting.
Turn what you saw into something the engineers can act on.
Check afterwards whether the fix actually fixed it.
Data analytics Work on the data that decides whether a model ships.: Every PSI engagement agrees what good looks like and then measures against it. That measurement is data work, and it is the part that decides whether a system is trusted. You would be inside it.; Clean and question the data behind live client systems.
Build the analysis that says whether a model is doing its job.
Spend more time on whether a number is trustworthy than on producing it.

Apply for an internship

Still studying is fine. Tell us what you are working on.

How to apply

Send us your CV and something you built.

Email [email protected] with the role in the subject line. A link to a repository, a paper or a product tells us more than a cover letter does, so send one if you have it.

Nothing here quite fits, but you think you should be working with us? Write anyway and say why.