Ivan Yakovlev
ivan [at] iyakovlev.dev · Barcelona
I am a Research Engineer at Gradium, where I work on real-time speech-to-speech translation, and data pipelines for ASR/TTS/S2ST. My research focuses on building production-grade speech models that scale across languages with limited supervision.
Before Gradium, I was a Senior Research Engineer at Palabra AI (2025–2026), working on multilingual TTS, streaming ASR, and speaker recognition. There I developed ReDimNet2.
Earlier, I was a Senior Machine Learning Engineer at ID R&D Inc. from 2019 to 2025, working on speaker verification. There I designed the ReDimNet architecture and contributed to our entries in the VoxCeleb and SASV challenges, which received top placements at Interspeech workshops.
I received my B.Sc. in Quantum Mechanics from Saint Petersburg State University in 2016.
news
| Sep 2026 | Presented ReDimNet2 (oral) at Interspeech 2026 in Sydney (arXiv, code) |
| Sep 2026 | Joined Gradium as Research Engineer |
| Mar 2026 | Released the ReDimNet2 preprint (arXiv) |
| Mar 2025 | Joined Palabra AI as Senior Research Engineer |
| Jul 2024 | Reshape Dimensions Network for Speaker Recognition accepted to Interspeech 2024 (arXiv) |
| Aug 2023 | 1st place (open track) at the VoxCeleb Speaker Recognition Challenge 2023; oral presentation at Interspeech 2023 |