Job Description

We are looking for an NLP Engineer.

Responsibilities:
  • Train and improve language models for the copilot. Adapt models to new conversation scenarios, response requirements and output formats
  • Prepare training and evaluation datasets. Analyse model errors, identify gaps in the data, define annotation tasks and review the resulting examples
  • Select models and develop prompts. Compare internally hosted models with external APIs in terms of quality, latency and cost
  • Maintain evaluation datasets and benchmarks. Check new versions for regressions and investigate poor responses and user feedback
  • Build services and APIs around models. Package and deploy them, add logging and monitoring, and investigate issues in production
  • Improve inference speed, GPU memory usage and serving cost while maintaining the required response quality
  • Take ownership of agreed tasks, keep experiments reproducible and document decisions and results so that other engineers can maintain the solution
Required Qualifications:
  • Strong Python skills and experience writing maintainable code beyond research notebooks
  • Practical experience with PyTorch and Hugging Face Transformers. Understanding of transformer models, tokenisation, training and text generation
  • Experience adapting language models through supervised fine tuning and approaches such as LoRA
  • Ability to prepare and validate datasets, prevent data leakage and investigate model errors
  • Experience evaluating generated text using benchmarks, human review and automated methods
  • Understanding of the limitations of each approach
  • Experience working with model APIs, prompts and structured outputs
  • Ability to build and maintain APIs using FastAPI or a comparable framework
  • Practical experience with Docker, Linux, Git and deploying models on GPUs. Familiarity with an inference framework such as vLLM or an equivalent tool
  • Ability to add logging, monitoring and appropriate tests, investigate production failures and roll back a faulty release
  • SQL skills sufficient to retrieve and analyse product data. English sufficient to read technical documentation and research papers
Benefits:
  • Vacation 28 calendar days per year
  • 7 wellness days per year (time off) that can be used to deal with household issues, to lie down and recover without taking sick leave
  • Children's education reimbursement. 50% compensation for the costs of kindergarten or school, up to $1000 gross per year per child, according to the paychecks after the integration period
  • Bonuses up to $5000 for recommending successful applicants for positions in the company
  • 50% payment for professional training, international conferences, and meetings
  • Corporate discount for English lessons
  • Health benefits. According to the paychecks, if you are not eligible for corporate medical insurance, the company will compensate you with up to $ 1,000 gross per year per employee. This can be spent on self-purchase of health insurance or on doctor’s fees for yourself and close relatives (spouse, children)
  • Workplace organization. The company provides all employees with an equipped workplace and all the necessary equipment (table, armchair, wifi, etc.) in our offices or co-working locations. In the other locations, the company provides reimbursement of workplace costs up to $ 1000 gross once every 3 years, according to the paychecks. This money can be spent on the rent of the co-working room, on equipping the working place at home (desk, chair, Internet, etc.) during those 3 years
  • Internal gamified gratitude system: receive bonuses from colleagues and exchange them for our merchandise, team building activities, massage certificates, etc.
Nice To Have:
  • Practical experience with recommendation systems, even if limited.
Technologies:
  • Python
  • PyTorch
  • Hugging Face Transformers
  • LoRA
  • FastAPI
  • Docker
  • Linux
  • Git
  • vLLM
  • SQL
Note:

✨ Our intelligent job search engine discovered this job and republished it for your convenience.
Please be aware that the job information may be incorrect or incomplete. The job announcement remains the property of its original publisher. To view the original job and its full details, please visit the job's URL on the owner’s page.