Specialist, AI Engineering

Merck Merck · Pharma · Telangana, India

Develops GenAI applications, including agentic AI systems, RAG pipelines, and multimodal solutions, using frameworks like LangChain and Google ADK. Focuses on prompt engineering, agent tooling, and integrating with enterprise services, with a minimum of 5 years of ML model development and deployment experience.

What you'd actually do

  1. Implement Agentic AI systems leveraging frameworks such as LangChain, Google ADK, Strands, and Semantic Kernel, enabling multi-step reasoning and task orchestration.
  2. Apply advanced prompt engineering techniques (ReAct, few-shot learning, chain-of-thought, prompt chaining) to improve accuracy, determinism, and robustness of LLM systems.
  3. Build multimodal AI solutions, integrating text with images or structured data to support complex enterprise use cases.
  4. Apply harness engineering practices to systematically design, test, and evaluate AI/agent workflows across different scenarios.
  5. Develop reusable agent tooling and integrations, enabling interaction with enterprise services, tools, and workflows, including Model Context Protocol (MCP) and Agent-to-Agent (A2A) communication patterns.

Skills

Required

  • Agentic AI systems
  • LangChain
  • Google ADK
  • Strands
  • Semantic Kernel
  • Prompt engineering techniques
  • ReAct
  • few-shot learning
  • chain-of-thought
  • prompt chaining
  • Multimodal AI solutions
  • Harness engineering practices
  • Agent tooling and integrations
  • Model Context Protocol (MCP)
  • Agent-to-Agent (A2A) communication
  • Scalable GenAI-driven solutions
  • Claude CLI
  • GitHub Copilot
  • Vertex AI
  • Codex
  • AWS
  • GCP
  • RAG architectures
  • embedding generation
  • retrieval optimization
  • grounding
  • context management
  • AgentOps practices
  • observability
  • tracing
  • logging
  • performance evaluation of AI systems
  • Python
  • tooling
  • function-calling
  • agent-driven integrations
  • NLP
  • text processing
  • tokenization
  • embeddings
  • preprocessing pipelines
  • transformer architectures
  • LLM ecosystems

Nice to have

  • Multimodal AI systems (vision)
  • LLM evaluation techniques
  • synthetic test generation
  • prompt regression testing
  • Advanced chunking strategies
  • context window optimization
  • TensorFlow
  • PyTorch
  • LLMOps
  • MLOps

What the JD emphasized

  • Minimum of 5 years of work experience in the end-to-end lifecycle of ML model development and deployment into production within a cloud infrastructure (Databricks is highly preferred).
  • Expertise in RAG architectures, including embedding generation, retrieval optimization, grounding, and context management.
  • Experience in prompt engineering techniques
  • Experience implementing AgentOps practices, including observability, tracing, logging, and performance evaluation of AI systems.
  • Strong Python programming skills with experience in tooling, function-calling, and agent-driven integrations.
  • Familiarity with harness engineering approaches for systematic testing, validation, and benchmarking of AI systems.
  • Understanding of Model Context Protocol (MCP) and Agent-to-Agent (A2A) interaction patterns.
  • Solid expertise in NLP and text processing, including tokenization, embeddings, and preprocessing pipelines.
  • Understanding of transformer architectures and modern LLM ecosystems.

Other signals

  • GenAI applications
  • Agentic AI systems
  • LLM-based workflows
  • multimodal AI solutions
  • scalable GenAI-driven solutions