AI Engineer - Voice

Teknologiens Mediehus; Stockholms kommun; Hybrid; SEK 60,000 - 90,000; Full time

Benefits offered by this job

Comprehensive range of employee benefits

High-trust team culture

Autonomy in role

Job summary

Teknologiens Mediehus is seeking a Voice AI Engineer to lead next-generation voice platform development. This hybrid role requires expertise in speech models and platform engineering.

The engineer will focus on STT and TTS for Nordic languages, implementing production-ready solutions. Candidates should have strong Python, experience in machine learning, and fluency in English and Nordic languages. Our team values fun and creativity, offering competitive benefits.

Qualifications

• Hands-on experience fine-tuning STT and/or TTS models.
• Experience taking models to production.
• Strong command of Python for both research and production use.

Responsibilities

• Drive STT and TTS quality across Nordic languages.
• Deploy next-generation speech-to-speech architectures.
• Design metrics and testing harnesses for speech model evaluation.

Skills

Hands-on experience with speech models

Strong Python skills

Experience with PyTorch

Fluency in English and one or more Nordic languages

Education

Degree in Computer Science, Machine Learning, or related field

Tools

PyTorch

Hugging Face

ONNX

Job description

Overview

Are you ready to be at the bleeding edge of Voice AI, building systems that handle millions of real conversations per month where every millisecond counts? At boost.ai, we are currently scaling up our Voice team. 2026 is the year of Voice! With a steady 15‑20% YoY growth rate at boost.ai, we are the go‑to Conversational AI provider for regulated sectors such as finance, the public sector, insurance, and telecom, and we are expanding into many more industries, including healthcare. We are a bona fide AI and Data Science company that has been pioneering machine learning since 2016. Today, we build a voice‑first platform where millions of conversations will run through, and more than 600 of our virtual agents are live across EMEA and North America. Enterprises like Tryg, DNB, Telenor, and Nordea trust us with their customers.

Challenge

The Challenge: The Nordic Gap & Zero Forgiveness. Voice is our biggest bet and our fastest‑growing product, but running speech models in a Jupyter notebook is entirely different from serving them live to millions of enterprise customers. Your mission is to tackle the exact ML problems that the major tech labs quietly ignore. You will take ownership of perfecting STT and TTS across the complex Nordic languages, curating datasets, fine‑tuning open and proprietary models, and defining the gold standard for accuracy in languages with limited open‑source coverage. Simultaneously, you will be taking next‑generation Speech‑to‑Speech (S2S) architectures straight from research papers into production. When interacting with an AI voice, users have zero forgiveness for unnatural pauses. You will be forced to constantly balance model accuracy against aggressive millisecond‑latency budgets and GPU economics. It is a fast‑moving, highly complex space where the research questions and the live production system are the same job.

Location

This role is hybrid and based in one of our European offices: Copenhagen, Oslo, Stavanger, Stockholm, Helsinki, or London. Eligibility to live and work in the Nordics or the UK is required. We are not able to support relocation from outside this region or fully remote arrangements for this position.

What you will be doing

You are a Voice AI Engineer who leads the next generation of our voice platform. The role sits right where applied ML meets platform engineering. Responsibilities include:

• Own Nordic language coverage: Drive STT and TTS quality across the Nordic languages, including underserved ones. Build dataset curation pipelines, fine‑tune open and proprietary speech models, and set the bar for quality.
• Bring speech‑to‑speech to production: Take next‑generation S2S architectures from prototype to production: latency‑aware, cost‑aware, and evaluation‑driven from day one.
• Build the evaluation harness: Design objective metrics, listening tests, and regression harnesses that allow confident swaps, tuning, and upgrades of speech models.
• Shape the inference architecture: Partner with platform engineering on model routing, multi‑region deployment, latency optimization, and CPU/GPU economics of real‑time speech.
• Stay at the frontier: Keep up with the fast‑moving world of speech and multimodal AI and bring advances onto our roadmap.
• Work across teams: Collaborate with Voice, ML, Core Platform and customer‑facing teams to turn language requirements and customer needs into shipped model work.

What we expect from you

We are open about seniority. Whether you are an experienced Applied ML Engineer or a Senior/Staff‑level practitioner with a real track record in speech, we want to talk.

Requirements (The Non‑Negotiables)

• Hands‑on experience with speech models: Fine‑tuning STT (Whisper, Parakeet, NeMo, or similar) and/or TTS (XTTS, VITS, StyleTTS, or commercial APIs).
• You have taken models all the way to production.
• Experience with PyTorch and modern inference stacks (Hugging Face, ONNX, Triton, or equivalent inference servers).
• Strong Python skills, comfortable across both research workflow and writing production‑grade code.
• A degree in Computer Science, Machine Learning, Computational Linguistics, Signal Processing, or a closely related field.
• Fluent in English and one or more Nordic languages.

Nice‑to‑haves

• You have designed both objective and subjective evaluation methods for speech quality; you know that what you can measure, you can improve.
• Comfortable deploying and operating on AWS, Azure, or GCP, familiar with GPU inference, model serving and latency‑critical workloads.
• Familiarity with Speech‑to‑Speech (S2S) architectures is a strong plus.

Team & Perks

We work hard on tough challenges, but fun is one of our actual core values. We are an opinionated, high‑trust team that celebrates wins together. You will have autonomy and trust to make a real difference to the direction of our voice technology. We offer a comprehensive range of employee benefits designed to support your overall well‑being and allow you to do the absolute best work of your career.

Contact

Ready to build the future of Voice AI? We would love to hear from you. If you have any questions about the role, the models we are working with, or the tech stack before applying, feel free to reach out to our recruitment partner: Casper Skyum Høgh, [e-mail], who is leading the search for this position.