AI Engineer

Job Description ManualAI Kindly share Immediate Joiners for the following position: Position Name Req ID Experience Location Duration Budget AI Engineer 723234 Senior | Senior Level 2 Chennai/Bangalore (India) 9 Months 25 USD/Hourly Support the development and integration of AI-driven solutions, build PoCs and automation workflows, and integrate LLM/AI services with enterprise applications. Required Skills Python, APIs, AI/ML integration, LLMs, prompt engineering, RAG frameworks, and Cloud AI services (Azure/OpenAI/AWS). Question Average Candidate Will Say Good Candidate Will Say Red Flags Tell me about a GenAI/LLM project you have built. Walk me through the architecture and your contribution. Explains the business use case and mentions using GPT/OpenAI. Can describe some components but lacks depth on architecture. Clearly explains end-to-end architecture, including data flow, APIs, retrieval layer, LLM integration, deployment, and personal contributions. Can justify design decisions. 🚩 Cannot explain architecture.🚩 Only discusses prompts or business requirements.🚩 Uses buzzwords without understanding.🚩 Contribution is unclear ("my team built it"). Strong AI Engineer Indicators ✅ Has built at least one LLM/RAG application end-to-end ✅ Comfortable with Python and API development ✅ Can explain embeddings, chunking, retrieval, and prompt engineering ✅ Has exposure to Azure OpenAI, OpenAI APIs, AWS Bedrock, or similar services ✅ Understands deployment, monitoring, and operationalization of AI systems ✅ Can discuss trade-offs and production challenges 2. If you need to build a chatbot over enterprise documents, how would you approach it? Mentions document upload, embeddings, and vector database at a high level. Understands the concept of RAG but lacks implementation details. Explains document ingestion, chunking strategy, embeddings, vector store, retrieval, metadata filtering, prompt augmentation, security/access control, and response grounding. 🚩 Suggests directly uploading documents to ChatGPT.🚩 Cannot explain RAG or embeddings.🚩 No understanding of retrieval mechanisms.🚩 No consideration for enterprise security. Immediate Concerns 🚩 Experience limited to using ChatGPT UI only 🚩 No hands-on coding experience with AI frameworks 🚩 Cannot explain architecture or data flow 🚩 No API integration experience 🚩 No understanding of RAG, embeddings, or vector databases 🚩 Cannot describe any real project ownership or production deployment 3. What frameworks have you used for AI orchestration (LangChain, LangGraph, Semantic Kernel, etc.)? Why did you choose them? Has used one or more frameworks and can explain basic workflow implementation. Explains orchestration patterns, agent workflows, tool calling, state management, multi-agent coordination, and reasons for selecting a specific framework over alternatives. 🚩 Knows framework names only.🚩 Cannot explain why it was selected.🚩 Cannot differentiate orchestration frameworks from LLM APIs.🚩 No practical usage. 4. Tell me about a challenge you faced while working with an LLM application and how you resolved it. Mentions common issues such as hallucinations or response quality and describes basic prompt tuning. Discusses a real production challenge (hallucinations, retrieval quality, latency, token limits, cost, deployment issues, etc.), root cause analysis, and measurable improvements achieved. 🚩 Claims there were no challenges.🚩 Gives generic blog-level answers.🚩 No troubleshooting experience.🚩 Cannot explain resolution approach. 5. How have you deployed or integrated AI applications in Azure, AWS, or enterprise environments? Has worked with cloud AI services and can describe basic deployment or API integration. Explains deployment architecture, CI/CD, secrets management, monitoring, logging, scaling, networking, security, and integration with enterprise applications. 🚩 Only local development experience.🚩 No cloud exposure.🚩 Cannot explain deployment lifecycle.🚩 No understanding of operationalization. 6. Explain a RAG implementation you worked on. How was the retrieval layer designed? "We stored documents in a vector database and searched them." Explains chunking strategy, embeddings, vector database, retrieval process, reranking, metadata filtering, evaluation metrics, and hallucination mitigation. Example: "We used Azure AI Search with metadata filters and hybrid search to improve precision." 🚩 Cannot explain embeddings. 🚩 Cannot explain chunking. 🚩 Thinks RAG is simply uploading PDFs to ChatGPT. 🚩 No retrieval strategy discussion.