Machine Learning Engineer, Agentic Marketing

Adobe Adobe · Enterprise · Bangalore, India

Machine Learning Engineer to build AI agents for marketing supply chain automation, focusing on planning, tool use, orchestration, and evaluation systems.

What you'd actually do

  1. Develop and build production agent systems on top of large language models, including planning, tool use and function calling, and long-running tasks that move marketing work through the pipeline.
  2. Own a core part of the agent stack from the first version to production. This could include orchestration and planning, the tool and function-calling layer, agent memory and state, or retrieval over brand and asset data.
  3. Build evaluation and reliability systems to ensure agent functionality, covering task success, grounding, regression checks, and brand-safety and compliance guardrails.
  4. Make architecture and build-versus-buy decisions, raising the engineering bar on the team.
  5. Collaborate closely with product, build, and research to turn current agent research into shipped features.

Skills

Required

  • Python
  • data structures
  • algorithms
  • distributed systems
  • API composition
  • building and shipping LLM-based applications in production
  • agentic systems on LLMs
  • multi-step planning
  • tool and function calling
  • orchestration
  • long-running tasks
  • evaluation systems for non-deterministic software

Nice to have

  • agent and orchestration frameworks such as LangGraph
  • LLM tooling such as vLLM, Hugging Face, and Weights & Biases
  • Fine-tuning or post-training experience (SFT, DPO, GRPO) applied to tool use or agent behavior
  • marketing, creative, or content technology
  • workflow-automation products
  • Open-source contributions
  • MS or PhD in Computer Science, Machine Learning, or a related field, or equivalent experience

What the JD emphasized

  • building and shipping LLM-based applications in production
  • Experience crafting agentic systems on LLMs
  • Experience building evaluation systems for software that does not behave deterministically

Other signals

  • AI agents
  • LLM applications
  • production systems
  • evaluation systems