Ai Ml Engineer

Role & responsibilities

1. Build and test LLM-based features using APIs (Claude, OpenAI, Gemini) prompt design, structured output, and evaluation of response quality.
2. Design and develop AI agents that plan, call tools, and complete multi-step business tasks with minimal human intervention.
3. Implement tool/function calling and integrate agents with internal REST APIs, databases, and enterprise systems (SAP, CRM, WMS) via MCP or equivalent connector patterns.
4. Build multi-agent and workflow-orchestration solutions using frameworks such as LangGraph, CrewAI, AutoGen, or the Claude/OpenAI Agents SDK including memory, state management, and error recovery.
5. Develop and maintain conversational flows for WhatsApp/web chatbots, including session handling and backend integration.
6. Work on Retrieval-Augmented Generation (RAG) use cases document ingestion, chunking, embeddings, and vector search over internal content.
7. Assist in developing natural-language-to-SQL and analytics assistants over MS SQL/enterprise databases with read-only guardrails.
8. Write Python scripts and REST APIs (Flask/ FastAPI) to expose AI features to internal applications.
9. Define guardrails, approval checkpoints, logging, and evaluation methods so agents behave safely and predictably in production.
10. Support data preparation cleaning, labelling, and structuring datasets for model evaluation.
11. Document solutions, prepare test cases, and present findings to business stakeholders.

Preferred candidate profile

1. Projects built with agent frameworks LangGraph, CrewAI, AutoGen, Semantic Kernel, or custom agent loops.
2. Awareness of the Model Context Protocol (MCP) and how agents connect to external tools and data sources.
3. Exposure to agent evaluation and observability tools (LangSmith, LangFuse, or custom tracing).
4. Academic or personal projects using RAG or vector databases (FAISS, Pinecone, Chroma, pgvector).
5. Exposure to Flask/FastAPI, Streamlit, or Node.js.
6. Basic understanding of ML concepts embeddings, classification, evaluation metrics. Published GitHub repositories, Kaggle work, or hackathon participation