Machine Learning Engineer - Model Training
About the job
- Experience training TTS or STT models directly and applying them in a service.
- Experience training speech models end-to-end using PyTorch.
- Deep understanding of Transformer-based speech model architectures.
- Experience building speech datasets and improving data quality.
- Experience training models in GPU environments, including multi-GPU setups.
- Experience deeply contributing to model performance improvement.
- Design and train proprietary AI models to solve product problems.
- Initially improve TTS/STT model performance and speech quality.
- Build end-to-end machine learning pipelines covering data collection, preprocessing, training, and evaluation.
- Quantitatively evaluate and improve speech quality, including naturalness, pronunciation accuracy, and emotional expression.
- Design efficient model architectures and serving methods that account for inference latency and cost.
- Collaborate with product engineers to apply research results to real services.
- Help shape technical direction and the research roadmap in a small team.
- Explore and research a range of AI model problems, including lightweight language models and routing models.
- Experience training multilingual speech models, such as Korean or Japanese.
- Experience building real-time streaming speech recognition or synthesis systems.
- Experience with distributed training, such as DeepSpeed or FSDP.
- Experience implementing or researching speech-related papers.
- Experience contributing to open-source speech models.
- Experience optimizing inference with tools such as ONNX or TensorRT.
- Experience researching lightweight language models or specialized models.
About the company
Channel Talk is an AI-powered customer-messaging and CRM platform that provides a website chat widget, a unified inbox for customer conversations, AI chatbots and automation, CRM marketing, and team messaging with integrations for channels like WhatsApp. Conversations from the chat widget feed into a unified inbox where AI bots can respond and automate tasks, while CRM tools manage campaigns and contacts and team messaging supports internal coordination. It stands out by pairing cross-channel messaging, automation, and CRM features in one platform with a regional focus on Southeast and East Asia, and it originated from Walk Insights analytics tools before shifting to customer-communication software. The goal is to help businesses improve customer engagement and support by centralizing conversations, automating routine tasks, and coordinating marketing and sales across channels.
Company Size
201-500
Company Stage
Late Stage VC
Total Funding
$43.7M
Headquarters
Seoul, South Korea
Founded
2014
Get referred to Channel Talk
See people who can refer or advise you
Simplify's Take
What believers are saying
- July 2026 Qoo10 integration expands distribution across major Japanese marketplace workflows.
- August 2026 Bカート partnership widens B2B commerce reach and ALF usage.
- January 2025 funding and 2026 voice improvements support Voice ALF commercialization.
What critics are saying
- Intercom Fin, Zendesk AI, and Gorgias AI attack Channel Talk's core support-automation budget.
- ALF phone remains beta in 2026, delaying enterprise migrations and expansion deals.
- Salesforce absorbing Intercom would squeeze pricing power and make Channel Talk a niche tool.
What makes Channel Talk unique
- ALF spans chat and voice, with IVR and task automation across channels.
- 2026 integrations with Qoo10 and Bカート deepen Channel Talk's EC workflow moat.
- 24万社超 adoption and 98% retention signal sticky SMB and midmarket usage.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Flexible Work Hours
Hybrid Work Options
Unlimited Paid Time Off
Health Insurance
Phone/Internet Stipend
Company News
Channel Corporation, developer of the AI chat service "Channel Talk" for EC and B2B, raised ¥1.1 billion, bringing total funding to ¥5.1 billion. The round was led by Bon Angels Venture Partners, with participation from Laguna Investment and Altos Ventures. "Channel Talk" is used by over 180,000 companies, boasting a 98% retention rate. The AI agent function "ALF" was released in November 2024, with 1,000 companies implementing it, and plans for "Voice ALF" are underway.
Channel Corporation, the developer of the AI chat service "Channel Talk" for EC and B2B services, announced on the 24th that it has raised ¥1.1 billion in a funding round led by existing investor Bon Angels Venture Partners, with participation from Laguna Investment and new investor Altos Ventures. This brings their total funding to ¥5.1 billion. The service is used by over 180,000 companies globally, with a 98% retention rate. The new AI agent feature "ALF" will be officially released in November 2024.
Channel Corporation, the operator of the all-in-one AI business messenger "Channel Talk," has raised ¥1.1 billion in new funding, bringing its total funding to ¥5.1 billion. The round was led by existing investor BonAngels Venture Partners, with participation from Laguna Investment and new investor Altos Ventures. The funds will be used for further AI development, including the "Voice ALF" feature. Channel Talk is expanding its AI capabilities to enhance customer support and plans to increase hiring.
Channel Corp, the company behind AI business messenger Channel Talk, raised $8.5M in funding, totaling $39.3M. The round was led by BonAngels Venture Partners, with Laguna Investment and Altos Ventures. The funds will develop Voice ALF, an AI voice solution, and support global expansion. Channel Corp operates in 22 countries, with 25% of revenue from global markets, and established a U.S. subsidiary in 2022. CEO Siwon Choi aims to position the company as a leading SaaS provider globally.
Channel Corporation announced on the 28th that it has launched an interactive voice response (IVR) function that allows users to design phone consultation scenarios without development knowledge on Internet phones provided by the all-in-one AI business messenger "Channel Talk."