Senior Software Engineer
6 - 10 years
San Francisco Bay Area
Onsite
1 month ago
About the role
About Lingopal
Live translation is one of the hardest real-time systems problems in media. Latency measured in milliseconds. Language nuance measured in culture. Audio fidelity measured in whether a broadcast feels human or broken.
Lingopal powers live multilingual broadcasts for major sports leagues, broadcasters, and enterprise media companies across the world. We separate audio from live streams, transcribe and translate it in real time, synthesize natural-sounding speech in the listener's native language, and reintegrate it seamlessly — at broadcast scale, with the fidelity that makes the difference between an experience that feels human and one that doesn't.
The Role
The hardest part of live translation isn't the translation. It's everything around it.
Pacing that feels natural in one language breaks down in another. Regional dialects — the difference between Mexican Spanish and Portuguese Spanish, the slang and phrasing that generic models flatten into something that sounds technically correct but culturally wrong. Audio recombination where translated voice and separated background audio have to be layered back together without distortion, without artifacts, without the seams showing.
These are the problems competitors haven't solved. They're the reason broadcasters choose Lingopal. And they're what you'll be working on.
As Senior Engineer on our Audio Pipeline team you sit at the intersection of audio DSP, real-time systems, and applied ML — not as a researcher, but as an engineer who ships. You understand the full pipeline from source separation to TTS mix-down, and you have the instincts to know where the quality is being lost and how to get it back.
What You'll Work On
What We're Looking For
Nice to Have
Why This Role
The gap between a translation that is technically correct and one that feels human is where this role lives. It's a hard problem — linguistically, acoustically, and technically. Most companies are stringing together third party APIs and hoping the result is good enough. We're not.
You'll work on the part of the pipeline that determines whether a global audience experiences a broadcast as immersive and natural or as obviously artificial. That's a problem worth solving, and the work you do here will be genuinely difficult to replicate.
What We Offer
We're establishing our Bay Area engineering headquarters between San Jose and San Francisco. This role is based in-office as we build that presence.
Live translation is one of the hardest real-time systems problems in media. Latency measured in milliseconds. Language nuance measured in culture. Audio fidelity measured in whether a broadcast feels human or broken.
Lingopal powers live multilingual broadcasts for major sports leagues, broadcasters, and enterprise media companies across the world. We separate audio from live streams, transcribe and translate it in real time, synthesize natural-sounding speech in the listener's native language, and reintegrate it seamlessly — at broadcast scale, with the fidelity that makes the difference between an experience that feels human and one that doesn't.
The Role
The hardest part of live translation isn't the translation. It's everything around it.
Pacing that feels natural in one language breaks down in another. Regional dialects — the difference between Mexican Spanish and Portuguese Spanish, the slang and phrasing that generic models flatten into something that sounds technically correct but culturally wrong. Audio recombination where translated voice and separated background audio have to be layered back together without distortion, without artifacts, without the seams showing.
These are the problems competitors haven't solved. They're the reason broadcasters choose Lingopal. And they're what you'll be working on.
As Senior Engineer on our Audio Pipeline team you sit at the intersection of audio DSP, real-time systems, and applied ML — not as a researcher, but as an engineer who ships. You understand the full pipeline from source separation to TTS mix-down, and you have the instincts to know where the quality is being lost and how to get it back.
What You'll Work On
- Audio source separation — advance our pipeline for cleanly separating voice from background audio in live broadcast streams, minimizing artifacts and preserving fidelity
- ASR/TTS pipeline — own and optimize the integration of transcription, translation, and speech synthesis services; reduce latency, improve accuracy, and eliminate the seams between stages
- Voice recombination and mix-down — solve the hard problem of layering translated TTS voice back onto separated background audio without distortion or unnatural pacing
- Regional language quality — work on the translation and synthesis quality problems that generic models get wrong; the dialect, slang, and phrasing differences that make the difference between correct and natural
- Pipeline latency — architect for the end-to-end latency targets that live broadcast demands; every stage you optimize compounds
- Observability and quality metrics — build the measurement systems that tell you objectively when audio quality improves or degrades across languages and broadcast conditions
- Competitive differentiation — contribute to the audio translation capabilities that competitors are actively trying to replicate and haven't
What We're Looking For
- 5+ years of backend systems engineering experience with depth in audio processing, real-time pipelines, or applied ML
- Experience with audio source separation — you understand the technical approaches, their tradeoffs, and where they break down in production
- Hands-on experience integrating ASR, translation, and TTS services into production pipelines — you've worked with these APIs at scale and know their failure modes
- Audio DSP fundamentals — sample rates, bit depth, mixing, resampling, and the signal processing concepts that underpin audio quality
- Strong understanding of pipeline latency — you can reason about end-to-end latency budgets and know which optimizations are worth making
- Experience building and operating real-time or latency-sensitive systems in production
- The instinct to measure quality objectively — you build evaluation frameworks, not just ship code and hope it sounds better
- Distributed systems fundamentals — you can operate and debug complex multi-stage pipelines at broadcast scale
Nice to Have
- Familiarity with Rust, Go, or C++ in a systems or audio processing context
- Experience at an audio or speech AI company — Deepgram, ElevenLabs, Resemble AI, Speechify, or similar
- Background in multilingual NLP or translation quality evaluation — understanding of how regional dialect and cultural context affect translation output
- Experience with FFmpeg, GStreamer, or similar media processing frameworks
- Familiarity with immersive or spatial audio formats
- Prior work on competitive audio differentiation — you've built something in this space that others are trying to catch up to
Why This Role
The gap between a translation that is technically correct and one that feels human is where this role lives. It's a hard problem — linguistically, acoustically, and technically. Most companies are stringing together third party APIs and hoping the result is good enough. We're not.
You'll work on the part of the pipeline that determines whether a global audience experiences a broadcast as immersive and natural or as obviously artificial. That's a problem worth solving, and the work you do here will be genuinely difficult to replicate.
What We Offer
- Competitive salary aligned with Series A benchmarks
- Meaningful equity — structured to reward long-term ownership and impact
- Full health, dental, and vision coverage
- PTO and Sick Leave
- Direct access to leadership and real influence over technical direction
- The chance to work on audio translation problems that define the category
We're establishing our Bay Area engineering headquarters between San Jose and San Francisco. This role is based in-office as we build that presence.