AI Engineer

AI Solution Design & Development

• Design, prototype, and validate AI-powered features spanning Generative AI, NLP, and Agentic AI use cases.
• Train, fine-tune, and evaluate language or vision models where pre-built or hosted models are insufficient — and operationalize them for inference in production.
• Architect and deliver production-ready, large-document advanced RAG workflows, including chunking strategies, hybrid retrieval, re-ranking, and evaluation.
• Build complex multi-agent systems — designing reusable, composable agent capabilities (skills, tools, actions) that can be dynamically invoked by LLMs.
• Implement agent interoperability and orchestration using protocols such as Agent-to-Agent (A2A), Agent Communication Protocol (ACP), and Model Context Protocol (MCP).
• Develop modular, reusable Python APIs and reference implementations for use cases including chatbots, document Q&A, summarization, and intelligent automation.
• Apply prompt engineering, context engineering, and solution tuning to optimize accuracy, latency, and cost.

Deployment & Optimization

• Provide well-documented proof-of-concepts and reference implementations to Full Stack and DevOps teams for integration and deployment.
• Collaborate with backend and cloud engineers to ensure AI solutions meet performance, cost, and security constraints.
• Build and optimize inference pipelines; monitor token usage, latency, and model performance, recommending improvements across the stack.

Product Innovation & Evangelism

• Act as an internal AI product evangelist — identifying, championing, and prototyping new AI-powered use cases.
• Collaborate with stakeholders to shape AI product concepts and contribute to roadmap development.
• Lead internal PoCs, technical demos, and feasibility assessments.
• Stay current with the evolving AI landscape and evaluate emerging tools, models, and techniques for adoption.

Required Skills & Experience

• 2–4 years in applied AI/ML engineering, with demonstrated delivery of production Generative and Agentic AI solutions.
• Ability to build AI solutions across Generative AI and Agentic AI, including training and fine-tuning language models when required and deploying them for inference.
• Proven experience building Agentic AI systems — multi-agent orchestration, reusable agent skills/tools, and agent interoperability (A2A, ACP, MCP).
• Hands-on experience designing and shippingadvanced RAG pipelinesin production — including hybrid retrieval, re-ranking, query transformation, and systematic evaluation — over large, unstructured document collections.
• Hands-on experiencebuilding and orchestrating agent skills— authoring and managing reusable skill definitions (e.g.,skills.md/Claude Skills-style capability files) that can be dynamically discovered and invoked by LLMs.
• Experiencebuilding custom MCP (Model Context Protocol) servers— exposing tools, resources, and data sources to LLMs/agents through standardized, interoperable interfaces.
• Proven experience buildingentity extraction and document understandingsolutions at scale across diverse, unstructured document formats.
• Practical experience with LLM orchestration frameworks and vector/retrieval systems (e.g., LangChain, LlamaIndex, LangGraph, CrewAI, AutoGen, Semantic Kernel; FAISS, Qdrant, Pinecone, Weaviate, Azure AI Search — or equivalents).
• Strong proficiency in Python, including API development (FastAPI/Flask), async/concurrent programming, data wrangling, and rapid prototyping (Streamlit, Gradio, etc.).
• Hands-on experience integrating at least one major LLM provider (e.g., OpenAI, Azure OpenAI, Anthropic/Claude, Hugging Face, Cohere, Meta Llama).
• Solid grasp of prompt engineering, context engineering, and latency-vs-cost trade-offs.
• Working familiarity with the Azure AI ecosystem (Azure AI Studio/Foundry, Azure OpenAI, Azure AI Services such as Vision, Translator, and Document Intelligence) — or willingness to ramp up quickly.
• Experience with the Anthropic/Claude ecosystem (Claude Skills, Claude Code, Agent SDK).
  • Familiarity with fine-tuning and model optimization techniques (PEFT, LoRA, quantization, distillation, pruning).