About this role
About SynhawkSynhawk builds omnimodal foundation models for communication integrity, aimed at infrastructure-side deployment in telco and banking sectors. Our platform analyzes the integrity of audio and video, and protects people against AI threats. This currently includes detecting synthetic speech & voice cloning, video & image manipulation, social engineering, and identity impersonation.
We're past the initial research phase. We're actively deploying with major telco customers, working with and building our own GPU clusters, air-gapped environments, and strict production SLAs. Your work with us will have an immediate, measurable impact on systems that defend real communications infrastructure.
We're a small, highly technical, founder-led team. You'll help make sure the models we ship hold up in the real world from day one.
The RoleWe spent a long time building these models. Now we'd like to pay someone to break them.
You'll sit between Synhawk Labs (R&D) and our BackEnd engineering team. You'll test our models in raw form and inside the apps we ship, work out why they break, and turn each failure into something the team can reproduce and fix. The work goes from big-picture questions ("what new use cases and data should these models handle?") down to small details: configurations, encodings, and how a file was recorded or re-encoded, which can all change a model's answer.
You don't need to be the most technical person in the room. We care more that you're curious, you use AI tools a lot, and you find unusual ways to break things.
What You'll DoTest our models end-to-end, both as raw models and inside shipped apps
Work out whether each failure comes from the model, the preprocessing pipeline, or the app
Write clear replication reports so the team can reproduce every weakness
Work with Labs and engineering on fixes, and suggest your own creative workarounds
Find new use cases and data distributions where our models could be used, and flag where they fall short
Get into the details: formats, codecs, encodings, configurations
Track findings across releases so we can see whether a class of weakness is actually getting fixed
Who We're Looking ForWhether you're early in your career, finishing your studies, or bringing years of industry experience, what matters most is that you spend a lot of time with AI models: testing them, breaking them, and working out how they behave. When something looks odd, you keep digging until you understand it.
You don't need to be a developer or an ML engineer. You should understand at a high level how machine learning models work and why the data they see affects how robust they are.
You'll fit well if you have:
A basic understanding of how ML models work and why data distribution matters
Some hands-on experience with development, ML, or data processing
A structured way of documenting what you find
Professional-level English, written and spoken
Bonus points for:
Adversarial thinking you can show us: CTFs, bug bounties, security research, jailbreaking or red-teaming models, QA on complex systems, or a portfolio of things you've broken (on purpose, ideally)
Experience with audio or video: recording, editing, formats, codecs
What We OfferA central role in a small team where your findings directly shape what ships
Close work with both our Prague engineers and our research lab in San Francisco
Real mentorship and room to grow fast
Flexible working hours and hybrid setup