We Build What Should Already Exist.

Speech Intelligence For Asian Accented Speech

600M people whose speech systems don’t understand.

Best use of VALSEA API at LotusHacks will receive a fast-track interview for the 9 open roles at VALSEA.

Top teams will be invited to interview directly with the VALSEA team.

In Southeast Asia, speech is messy: accents shift, languages mix, meaning depends on context.

Existing models were never built for this.

We’re building the speech intelligence layer that makes real-world Asian speech actually understandable.

You will work on undefined problems in messy, real-world systems where edge cases are the norm.

You’ll be trusted to figure things out, own outcomes, and ship fast. We learn by building, observing, and improving.

We care about accuracy, reliability, and what actually works in production.

How We Operate

Not tasks. Not effort. Outcomes. Trust is not earned here — it is given on day one. You keep it by shipping.

If you see a problem, it is yours. No one needs to assign it. No one needs to approve it. You figure it out.

We assume you are acting in the best interest of the team and the product. Until proven otherwise. That is the default.

Trust is not earned here. It is given on day one. You keep it by doing what you said you would do, by when you said you would do it.

Don’t walk in front of me; I may not follow. Don’t walk behind me; I may not lead. Walk beside me and be my friend.

~ Albert Camus

Val (Founder, VALSEA)