ML Engineer

Industry: Technology solutions - consumer electronics, software, streaming

Client: global enterpise, Swedish entity

Work model: full time, hybird 3 days/week on-site from Warsaw (Rondo Daszyńskiego) or Krakow (Equal Business Park) office

Project language: English

Project length: ASAP - September 30th 2026

Onboarding: 2 weeks in Malmo, Sweden

Start: ASAP or 1 week notice ideally / at latest start August 3rd 2026

Assignment type: B2B

Remuneration: up to 220 PLN/h net + VAT

Summary: The ML Engineer role is primarily focused on the end-to-end development of machine learning models, encompassing data ingestion, model training, evaluation, and deployment processes. This role is crucial in enhancing the organization’s machine learning capabilities and delivering improved models for various applications.

Responsibilities:

• Enhance encoder models to boost model capacity and resolution.
• Develop multi-classification support for diverse map-feature attributes by establishing a complete data pipeline, training models, and conducting evaluations.
• Port a large-scale inference pipeline to C, and develop a CLI along with a frontend for job management, as well as create a plugin interface for results visualization.
• Migrate CoreML inference to MLX for cross-OS model execution.
• Train a confidence and ranking model based on outputs from existing autoregressive models.
• Implement curriculum training by ranking tasks based on difficulty and automating the progressive scheduling of training data.
  
  

Key Requirements:

• Proficient in Python with experience in the full model development lifecycle (data pipelines, training, evaluation, deployment).
• Strong understanding of encoder architectures, autoregressive models, or self-supervised pre-training methodologies.
• Familiarity with CoreML, MLX, or similar inference frameworks.
• Experience in building data pipelines through to model serving and frontend integration.
  
  

Nice to Have:

• Proficiency in C or C++ for inference porting tasks.
• Knowledge of curriculum learning or training-data scheduling techniques.