Research Engineer, Forge

Jobtailor; Paris; Sur place; EUR 70 000 - 100 000; Plein temps

Résumé du poste

Jobtailor is seeking an experienced ML Engineer to translate real customer requirements into reliable training and deployment workflows, spanning model adaptation, post-training evaluation, data pipelines, and infrastructure.

You will build and improve post-training and evaluation workflows for CPT, SFT, RL, and distillation, develop tools for synthetic data generation, debugging, observability, and scalable systems across cloud and on‑prem environments.

Qualifications

• Experience with large codebases, testing, CI, and deployment ownership.
• Hands-on experience with LLM training or post-training pipelines.
• Ability to debug distributed ML jobs and data issues.
• Strong communication with technical and non-technical stakeholders.
• Experience building ML infrastructure and data pipelines.

Responsabilités

• Turn real customer requirements into reliable training and deployment workflows
• Work end-to-end across model adaptation and post-training, evaluation, data, and infrastructure
• Build and improve post-training and evaluation workflows for CPT, SFT, RL, and distillation
• Develop tools and pipelines for synthetic data generation, data curation, training, evaluation, and deployment
• Debug and harden large-scale ML systems, including distributed training, scheduling/execution, checkpointing, observability, and reproducibility
• Improve the Forge codebase through clear APIs, tests, documentation, and maintainable abstractions
• Advance the RL training stack, including high-throughput asynchronous rollout and scalable post-training systems
• Ensure Forge deployment is seamless and adaptable across diverse clients, hardware, software stacks, cloud, and on-premises environments
• Partner with researchers and infrastructure engineers to translate bottlenecks into concrete system improvements
• Collaborate with scientists, engineers, product, and customer-facing teams to ship maintainable and trusted Forge projects

Connaissances

Python Engineering

PyTorch

JAX

Distributed Systems

Data Pipelines

Testing & CI

Operational Ownership

Clear Communication

Outils

FSDP

DeepSpeed

Megatron

SLURM

Ray

Kubernetes

Kueue

Karpenter

Skypilot

Description du poste

• Turn real customer requirements into reliable training and deployment workflows
• Work end-to-end across model adaptation and post-training, evaluation, data, and infrastructure
• Build and improve post-training and evaluation workflows for CPT, SFT, RL, and distillation
• Develop tools and pipelines for synthetic data generation, data curation, training, evaluation, and deployment
• Debug and harden large-scale ML systems, including distributed training, scheduling/execution, checkpointing, observability, and reproducibility
• Improve the Forge codebase through clear APIs, tests, documentation, and maintainable abstractions
• Advance the RL training stack, including high-throughput asynchronous rollout and scalable post-training systems
• Ensure Forge deployment is seamless and adaptable across diverse clients, hardware, software stacks, cloud, and on-premises environments
• Partner with researchers and infrastructure engineers to translate bottlenecks into concrete system improvements
• Collaborate with scientists, engineers, product, and customer-facing teams to ship maintainable and trusted Forge projects

Requirements

• Strong Python engineering skills and experience working in large codebases, including testing, code review, CI, and operational ownership
• Hands-on experience with PyTorch, JAX, or similar
• Strong systems and infrastructure fundamentals
• Experience with LLM training or post-training, including fine-tuning, RL, distillation, evaluation, and/or data pipelines
• Excellent debugging skills in distributed jobs, data issues, quality regressions, and infrastructure failures
• Clear communication with technical and non-technical stakeholders
• High agency, low ego, and comfort in fast-moving, under-specified environments
• Nice to have: distributed training experience with FSDP, DeepSpeed, Megatron, or similar
• Nice to have: cluster/orchestration experience with SLURM, Ray, Kubernetes, Kueue, Karpenter, Skypilot, or similar
• Nice to have: experience building reliable ML infrastructure, evaluation systems, or large-scale data processing pipelines
• Nice to have: research experience in LLMs, agents, multimodal models, reasoning, code, or domain adaptation
• Nice to have: open-source contributions, publications, or widely used internal tooling
• Nice to have: experience training multi-billion-parameter models and on petabyte- and exabyte-scale datasets
• Nice to have: ability to identify bottlenecks across the stack and drive improvements from first principles

Core Competencies

Demonstrates strong Python engineering skills and hands-on experience with PyTorch or JAX, focusing on building and improving ML workflows, including training, evaluation, and deployment. Capable of collaborating with cross-functional teams to enhance system performance and reliability in fast-paced environments.

Highest-signal resume keywords

• Python Engineering
• ML Workflow Development
• Distributed Training
• Post-Training Evaluation
• Clear Communication

ATS Optimization Keywords

Hard Skills

• Python
• PyTorch
• JAX
• LLM Training
• Debugging
• Data Pipelines
• Testing
• Code Review
• Operational Ownership
• Infrastructure Fundamentals

Soft Skills

• Clear Communication
• High Agency
• Low Ego

Industry Keywords

• Model Adaptation
• Synthetic Data Generation
• Large-Scale ML Systems
• Observability
• Reproducibility
• Maintainable Abstractions
• Scalable Systems
• Bottleneck Identification

Tools & Technologies

• FSDP
• DeepSpeed
• Megatron
• SLURM
• Ray
• Kubernetes
• Kueue
• Karpenter
• Skypilot