Research · Paid role

Research Scientist, AI

Lead the research agenda behind Olyxee Document Integrity: verification, evaluation, and reliability for AI systems that take action in production.

ResearchHybrid · JohannesburgStaff · 8+ years of applied research

Compensation

Top-of-band salary, significant equity, performance bonus, conference and learning budget.

What you'll own

Set and own the multi-year research agenda on model reasoning, verification, and reliability
Design experiments, build evaluation harnesses, and publish work the field builds on
Translate research into product capabilities the engineering team can ship
Represent Olyxee in the research community through papers, talks, and collaborations
Mentor researchers and engineers and raise the technical bar across the company

Who we're looking for

PhD in CS, ML, or a closely related field, plus 8+ years of research beyond your doctorate
Sustained first-author record at top venues (NeurIPS, ICML, ICLR, ACL) with a public Google Scholar profile
Demonstrable citation impact, with work others build on and cite
Expert-level Python and modern ML frameworks (PyTorch, JAX); you write research- and production-grade code
Experience setting and owning a research agenda end to end, not just executing one
Patents, widely used open-source, or shipped research that reached real users
Exceptional written communication; you can make complex ideas land for any audience

How we'll evaluate you

01
Written application
Submit the full application below. We read every word. Expect a response within 14 days, even if it is a no.
02
Founder screen
A 30 minute conversation with the founder. We talk about your trajectory, your work, and how you think.
03
Take-home exercise
A paid, role-specific exercise that takes 6 to 10 hours over a week. We pay market rate for your time on this.
04
Technical deep dive
Two hours with two people from the team. We go deep on your exercise, your past work, and a live problem in your domain.
05
Final interviews
Three to four conversations with people you would work with most closely. We make sure both sides have everything they need.
06
References and offer
We contact your references, then move quickly to a written offer with compensation, equity, and start details spelled out.

Apply

The bar is high. Take your time.

Plan for 30 to 45 minutes. Strong applications are specific, written in your own voice, and show real proof of work. Vague answers are the most common reason we say no.

First name *

Surname *

Email *

Phone number *

Where are you based? *

LinkedIn profile *

GitHub, portfolio, or other relevant work *

Google Scholar, arXiv, or publications profile

Required for research, ML, and data science roles. Leave blank if it does not apply.

Relevant certifications or credentials

List any professional certifications relevant to this role.

Link to your CV *

Make sure the link is publicly viewable.

Current role and company *

Years of relevant experience *

Describe a project you led end to end that you are most proud of *

Be specific. Vague answers are the most common reason we say no.

Why this role and why Olyxee? *

Describe the hardest technical or strategic problem you have solved in the last 12 months *

Two professional references (name, relationship, email) *

Salary expectation (annual, in your local currency) *

Earliest start date *

Work authorization *

I confirm that everything in this application is accurate and written by me. I understand Olyxee will verify references and may conduct background checks.

Your application is sent directly to our hiring team. No third parties.