Machine Learning Engineer

Train and ship the speech, language, and conversation models that decide whether enterprise customers trust us with their calls.

About OneInbox

We're putting voice at the heart of how enterprises actually run.
We're the production-grade voice AI platform built for Japan and Asia-Pacific.
We're handling millions of real calls every month — in Japanese, English, Hindi, and Arabic — for banks, real estate firms, and government services.
Try OneInbox today.

Voice AI in regulated enterprise contexts lives or dies on real production metrics — not benchmark scores.
We have proprietary call data across multiple languages and verticals — and not enough hands on it.
The next 12 months of model improvements directly determine our ceiling on enterprise expansion.

Get hands-on with the current speech and conversation stack — ASR, TTS, intent, LLM orchestration.
Set up your evaluation harness against real production call data, not synthetic benchmarks.
Identify the 3 highest-leverage model improvements based on customer-visible metrics.

Ship a measurable improvement to one production-critical model.
Define and own the evaluation pipeline that the rest of the team will trust.
Reduce inference latency or cost on at least one hot path without dropping accuracy.

Ship a second model improvement, this time in collaboration with engineering and product.
Stand up a continuous training and eval loop that runs unattended.
Document the model surface so the next ML hire ramps in half the time.

3+ years of applied ML — speech, NLP, or LLM specialization preferred.
Comfortable with PyTorch, Hugging Face, and modern training infrastructure.
You've shipped ML to production and felt the difference between offline and online metrics.
Strong instinct for when to fine-tune, when to retrieve, and when to engineer the prompt.
Bonus: experience with multilingual or low-resource language modeling.

Generational impact: build the voice layer that every APAC enterprise eventually runs on.
Ownership culture: you'll work close to the founders and ship in days, not quarters.
Kind, high-bar team: senior operators who've built and led at scaled companies — and are easy to work with.
Strong backing: institutional investors and committed enterprise customers from day one.

Real ownership: competitive salary and meaningful equity for early team members.
Health coverage: medical, dental, and vision for you and your dependents.
Team time: regular in-person off-sites across Tokyo, Singapore, Dubai, and Bengaluru.
Flexible time off: take what you need, when you need it.
Learning budget: dedicated annual L&D, plus conference and travel support.
Daily essentials: catered meals on office days, commuter support, and wellness benefits.