Leads the team this role joins
Runs the team day to day
Decides on early hires
From your own Gmail and LinkedIn.
About the role
This is a full-time, fully remote NLP engineering role open to candidates worldwide, working on language models for a conversation copilot at Social Discovery Group. It suits engineers who can train and evaluate language models in PyTorch and turn them into maintainable, monitored APIs. The group runs social discovery and dating products across many brands and regions. The team works remotely from many countries.
What you’ll do
- Train and improve language models for the copilot, adapting them to new conversation scenarios and output formats.
- Prepare training and evaluation datasets, analyse model errors, and define annotation tasks.
- Choose models and write prompts, comparing self-hosted models with external APIs on quality, latency, and cost.
- Maintain evaluation sets and benchmarks, and check new model versions for regressions.
- Build and deploy model services and APIs with logging and monitoring.
- Improve inference speed, GPU memory use, and serving cost without losing response quality.
- Document experiments and decisions so other engineers can maintain the work.
What they’re looking for
- Strong Python skills and code written to production standards, not only in notebooks.
- Practical PyTorch and Hugging Face Transformers experience, including transformer architecture, tokenization, training, and text generation.
- Experience with supervised fine-tuning and methods such as LoRA.
- Ability to prepare and validate datasets, prevent data leakage, and diagnose model errors.
- Experience evaluating generated text with benchmarks, human review, and automated methods.
- Experience working with model APIs, prompts, and structured outputs.
- Ability to build and maintain APIs with FastAPI or a similar framework.
- Practical experience with Docker, Linux, Git, and GPU deployment.
- Familiarity with an inference framework such as vLLM or an equivalent tool.
- Ability to add logging, monitoring, and tests, and to investigate and roll back faulty production releases.
- SQL skills sufficient to retrieve and analyse product data.
- English good enough to read technical documentation and research papers.
Nice to have
- Practical experience with recommendation systems, even if limited.
What’s on offer
- Fully remote, full-time work.
- Pay range shared during recruitment, before employment starts.
- 28 calendar days of vacation per year and 7 wellness days per year.
- Health compensation of up to 1,000 USD gross per year where corporate medical insurance is not available.
- Workplace setup reimbursement of up to 1,000 USD gross once every 3 years for employees in locations without a company office.
- Reimbursement of 50 percent of kindergarten or school costs, up to 1,000 USD gross per child per year.
- Referral bonuses of up to 5,000 USD.
- 50 percent funding for professional training, international conferences, and meetings.
- Corporate discount on English lessons and an internal peer recognition system.
Skills
Questions about this role
Can I work from anywhere?
Yes, the role is remote and open worldwide.
How much does it pay?
The posting does not state a figure. The pay range is shared during recruitment, before employment starts.
How much vacation is included?
The role includes 28 calendar days of vacation per year and 7 wellness days per year.