Senior ML Engineer - Firefly

Adobe Adobe · Enterprise · Bucharest, Romania

Senior ML Engineer at Adobe Firefly team, focusing on adapting large-scale pretrained multimodal generative models (image, text, video) for creative products. Responsibilities include research, prototyping, integrating new techniques, building end-to-end pipelines, and contributing to data strategy and user workflows. Requires strong ML engineering, deep learning, generative model, and Python skills, with experience in large-scale ML pipelines from training to inference.

What you'd actually do

  1. Engage with large-scale pretrained models for multimodal generation (image, text, video, and beyond) and tailor them to support Adobe Firefly products and features.
  2. Research, prototype, and integrate brand-new techniques from the latest literature into production-grade generation pipelines.
  3. Construct and refine innovative end-to-end pipelines for multimodal generation, balancing quality, performance, and scalability.
  4. Work alongside data and research teams to review, modify, and recommend improvements to data strategies that support model training and fine-tuning.
  5. Analyze and iterate on user-facing workflows, translating product requirements into model and pipeline build decisions.

Skills

Required

  • 5+ years of industry experience in machine learning engineering
  • deep learning for computer vision or multimodal generation
  • handling extensive, intricate ML codebases
  • current ML frameworks like PyTorch or Tensorflow
  • constructing and refining ML pipelines on a large scale
  • model training and fine-tuning
  • inference and deployment
  • generative model architectures (e.g., diffusion models, transformers, GANs, VAEs)
  • adapting or training them for applied use cases
  • Strong Python proficiency
  • software engineering fundamentals
  • system build, testing, and debugging
  • distributed training or inference on GPU clusters
  • extensive knowledge of multimodal generation workflows
  • image or video generation
  • collaborating cross-functionally with research scientists, product managers, and infrastructure engineers
  • communicate technical trade-offs clearly
  • contribute meaningfully to roadmap discussions and architectural decisions
  • Comfort with ambiguity
  • thrive in a fast paced research and product environment
  • contribute to technical documentation, development proposals, or engineering standards

Nice to have

  • controllable generation
  • LoRA / fine-tuning workflows
  • RLHF-based alignment techniques
  • cloud platforms (AWS, Azure, or GCP)
  • MLOps tooling
  • creative AI or media technology domain
  • Contributions to open-source ML projects
  • peer-reviewed publications
  • Slurm or similar
  • Master's degree or PhD

What the JD emphasized

  • 5+ years of industry experience in machine learning engineering
  • extensive, intricate ML codebases
  • constructing and refining ML pipelines on a large scale — covering model training and fine-tuning all the way through inference and deployment
  • extensive knowledge of multimodal generation workflows

Other signals

  • adapting foundational models
  • production-grade generation pipelines
  • multimodal generation
  • large-scale pretrained models