PROACHIVE HIRING***The customer is looking for a Software Engineer who can contribute to the design and development of AI-driven applications using Java, Python, and modern Gen AI frameworks. You'll work on prompt engineering, Retrieval-Augmented Generation (RAG), agentic workflows, and scalable backend systems that integrate with LLMs and enterprise data using the Model Context Protocol (MCP).Key Responsibilities:
Develop and maintain backend services and AI agents using Java and/or Python.
Build and optimize Gen AI applications using:- Prompt engineering- Retrieval-Augmented Generation (RAG)- Agentic flows (multi-agent orchestration, tool use, memory)
Implement Model Context Protocol (MCP) to standardize AI interactions with external tools, databases, and services.
Knowledge with vector databases (e.g., FAISS, Pinecone, Weaviate) for semantic search and context retrieval.
Collaborate with product and research teams to translate AI roadmap goals into technical deliverables.Required Skills:
Strong programming skills in Java and/or Python.
Solid understanding of Generative AI concepts, including:- Prompt design and tuning- RAG pipelines and document chunking- Agentic workflows (e.g., LangChain agents, tool calling)
Familiarity with LLM APIs (OpenAI, Anthropic, etc.) and open-source models e.g., LLaMA, Mistral.
Experience implementing MCP for scalable, secure, and standardized AI-to-tool integrations.
Knowledge of RESTful APIs, microservices, and cloud platforms (AWS, Azure, GCP).Nice to Have:
Experience with Gen AI frameworks like LangChain, LlamaIndex, or Haystack.
Exposure to embedding models, tokenization, and context window optimization.
Understanding of MCP architecture: clients, servers, and host applications.
Prior work in AI/ML projects, NLP, or conversational AI.
Knowledge of CI/CD pipelines, containerization (Docker, Kubernetes), and observability tools.
Knowledge on integrating LLMs (e.g., OpenAI, Anthropic, Mistral) with internal systems and APIs.Relevant Gen AI Topics You'll Work With:
Prompt engineering best practices (zero-shot, few-shot, chain-of-thought)
Retrieval-Augmented Generation (RAG) with hybrid search
