Applied ML Engineer: From Research to Production

Lever, Inc.; France; Sur place; EUR 23 000 - 36 000; Plein temps

Résumé du poste

Lever, Inc. on behalf of a partner company is seeking an Applied ML Engineer in Romania to bridge machine learning research with production systems. You will design experiments, evaluate models, and build tooling for repeatable verification and reporting.

You’ll work across model internals, infrastructure, and user-facing dashboards, contributing to production-grade systems with APIs and observability. A strong Python, PyTorch, and open-weight model background is essential.

Qualifications

• Strong Python engineering skills with PyTorch experience.
• Experience with ML evaluation and experiment design.
• Ability to implement methods from first principles.
• Professional software engineering beyond notebooks.
• Familiarity with open-weight models and LLM inference systems.
• Experience with React/TypeScript is a plus for frontend visibility.

Responsabilités

• Reproduce and evaluate machine learning research methods.
• Design evaluation datasets, probes, scoring approaches, and baselines.
• Work with model weights, logits, hidden states, and inference infrastructure.
• Build and extend evaluation infrastructure for experiments and reports.
• Translate research workflows into intuitive product experiences and dashboards.
• Run controlled experiments comparing base, fine-tuned, and quantized models.

Connaissances

Python

PyTorch

Hugging Face Transformers

API design

Asynchronous jobs

Databases

Testing

Documentation

React/TypeScript knowledge

Analytical judgment

Outils

DSPy

LiteLLM

Temporal

Ray

vLLM

PostgreSQL/pgvector

Next.js

React

TypeScript

Description du poste

Lever, Inc. on behalf of a partner company is seeking an Applied ML Engineer in Romania to bridge machine learning research with production systems. You will design experiments, evaluate models, and build tooling for repeatable verification and reporting.

You’ll work across model internals, infrastructure, and user-facing dashboards, contributing to production-grade systems with APIs and observability. A strong Python, PyTorch, and open-weight model background is essential.