Remote Senior ML Engineer – LLM & Agent Systems Lead

Dating Group; Polska; A distancia; PLN 180.000 - 340.000; Jornada completa

Ventajas ofrecidas por este puesto de trabajo

Remote opportunity

Vacation 28 days/year

Wellness days 7/year

Bonuses up to $5000 for referrals

Training & conference support 50%

Health benefits up to $1000 gross/year

Workspace reimbursement

Merchandise & team events

Descripción de la vacante

Social Discovery Group (SDG) is hiring a Machine Learning Engineer to lead the ML direction, select strategic bets, and drive end-to-end LLM post-training and adaptation. You will design agent harnesses, build multi-agent workflows, and ship production systems shaping user experience.

Ideal candidates have hands-on experience with LLM fine-tuning (SFT, LoRA/QLoRA, PEFT), dataset construction, LLM evaluation, tool calling, and MLOps.

Formación

• Experience in training and fine-tuning LLMs.
• Experience with SFT/LoRA/QLoRA/PEFT methods.
• Ability to build datasets for LLM training.

Responsabilidades

• Set the ML direction and tie bets to business needs.
• Drive LLM post-training and model adaptation end to end.
• Design and build agent harnesses and tool-using LLM systems.
• Shape multi-agent development workflows to improve productivity.
• Build evaluation frameworks to predict real-world performance.
• Improve dialogue quality across memory and safety.

Conocimientos

LLM fine-tuning

Dataset construction

LLM evaluation

Agent harness design

Production LLM integration

Observability

Transformer architecture

Git

Docker

Linux

MLOps

Russian language

Herramientas

SFT/LoRA/QLoRA/PEFT

Descripción del empleo

Social Discovery Group (SDG) is hiring a Machine Learning Engineer to lead the ML direction, select strategic bets, and drive end-to-end LLM post-training and adaptation. You will design agent harnesses, build multi-agent workflows, and ship production systems shaping user experience.

Ideal candidates have hands-on experience with LLM fine-tuning (SFT, LoRA/QLoRA, PEFT), dataset construction, LLM evaluation, tool calling, and MLOps.