Agentic AI Engineer

We're looking for a hands-on Agentic AI Engineer to build, optimize, and deploy production-grade AI solutions. In this role, you will be the engine room of our AI initiatives, taking architectural blueprints and turning them into scalable, functional systems. You will work on initiatives, full build lifecycle from prototyping to production, collaborating with Senior Architects and data scientists to deliver AI solutions in a consulting environment.

The core responsibilities for the job include the following:

Agentic Development:

• Build: Develop LLM-based applications and multi-step agentic workflows using frameworks such as Microsoft Agent Framework, AutoGen, LangChain, LangGraph, LlamaIndex, or CrewAI.
• RAG Pipelines: Implement Retrieval-Augmented Generation pipelines: chunking, embedding, vector search, and re-ranking.
• Tool Use and Memory: Build agents with tool-calling, short/long-term memory, and human-in-the-loop checkpoints.
• Prompt Engineering: Design and iterate on system prompts, chain-of-thought templates, and structured output schemas.
• Model Fine-tuning: Execute fine-tuning and optimization tasks (quantization, PEFT/LoRA) to adapt foundation models for specific domain tasks.

Integration and Delivery:

• APIs: Expose AI capabilities via FastAPI endpoints; integrate with client data sources and third-party APIs.
• Vector Databases: Manage embeddings and retrieval using Pinecone, Qdrant, or pgvector.

MLOps and Productionization:

• Deployment: Containerize and deploy AI services using Docker and Kubernetes on AWS / Azure cloud environments, ensuring high availability and low latency.
• Observability: Implement observability for AI systems using LangSmith or Arize Phoenix, tracking accuracy, hallucinations, cost, and latency.
• CI/CD: Maintain CI/CD pipelines for ML, ensuring automated testing (unit, contract, and model-quality tests) is integrated into the delivery workflow.
• Guardrails and Hallucination Control: Apply output validation, guardrails, and hallucination-detection techniques to ensure reliable, production-safe AI outputs.
• Token Optimization: Apply prompt compression, context window management, and response caching to control inference cost and latency.

Collaboration:

• Client Delivery: Work closely with Senior Architects and Engagement Managers to translate business requirements into technical tasks and working solutions.
• Code Quality: Write clean, modular Python; participate in peer code reviews and contribute to the team's internal library of reusable AI patterns and playbooks.

Requirements:

• Experience: 2-5 years in software engineering, data engineering, or ML; at least 1 year building LLM/Gen AI applications.
• Python: Strong Python skills in OOP, async programming, packaging, and testing.
• LLM Frameworks: Hands-on experience with at least one of LangChain, LangGraph, CrewAI, AutoGen, or LlamaIndex.
• Gen AI: Working knowledge of LLM APIs (OpenAI, Anthropic Claude, Gemini) and prompt design.
• Cloud Basics: Familiarity with AWS or Azure; comfortable with REST APIs and Git-based workflows.
• MLOps and Productionization: Hands-on experience deploying, monitoring, and maintaining AI systems in production. Docker/Kubernetes, CI/CD pipelines, and observability tooling are non-negotiable.
• Engineering Fundamentals: Solid SQL skills, API design experience (FastAPI/Flask), and a software engineering-first approach to ML encompassing testing, modularity, and documentation.
• Education: B. Tech / B. E. / M. Sc. in Computer Science, Information Technology, or a related field.

Good-to-Have:

• Evaluation: Exposure to G-Eval, RAGAS, TruLens, or LangSmith for quantifying LLM output quality.
• MLOps: Basic experience with MLflow or DVC for experiment tracking.
• Structured Outputs: Experience with Pydantic-based output parsing and function/tool calling.
• Performance Tuning: Knowledge of vLLM or Triton Inference Server for high-throughput model serving.
• Certifications: AWS / Azure AI Fundamentals or equivalent cloud certification.