Job Title: AI Engineer - Agentic & GenAI Systems

Experience: 5-8 years

Location: Pune, India

Work Model: Hybrid (3 days Work From Office, 2 days Work From Home)

Must Have: JavakubernetCI/CDDocker,Ai Ml Engineer,Python,Large Language Model,Retrieval

Tech Stack & Qualifications

Primary Language: Python is mandatory. Knowledge of Java, Go, or Node.js is considered optional.
Frameworks: LangGraph or Semantic Kernel. FastAPI is a good to have.
Cloud Platforms: Experience in any one major platform (Azure, AWS, or GCP).
Operations (Optional): Packaging services as containers and deploying to Kubernetes with Helm/Argo CD is a bonus but not mandatory. Platform-level concerns like tenant isolation are handled by a separate team

Agent & Application Engineering

Multi-Agent Systems (MAS): Design systems involving planning, tool-use, and delegation using frameworks like LangGraph or Semantic Kernel.
Model Context Protocol (MCP): Integrate tools, SQL, search, and document stores using MCP with strict type contracts and safe sandboxes. Focus is on the consumption of MCP servers rather than building them from scratch.
Model Gateway Integration: Integrate with model gateways (OpenAI, Azure OpenAI, Bedrock, or Vertex AI). Knowledge of at least one of these platforms is sufficient.
Pro-Code Development: Build solutions using high-level coding (Python-focused); low-code/no-code experience alone is not sufficient.

Retrieval, Data & Knowledge

RAG Services: Stand up Retrieval-Augmented Generation services, including chunking, enrichment, embeddings, and indexing using hybrid/vector search (e.g., pgvector, Pinecone, Weaviate, OpenSearch).
Ingestion Pipelines: Implement ingestion pipelines for diverse data sources like documentation, tickets, and CRM data using Airflow, Prefect, or Ray.
Optimization: Continuously optimize retrieval quality through chunking strategies, re-rankers, and query rewriting based on evaluation metrics.

Quality, Testing & Evaluation

Evaluation Frameworks: Utilize Promptfoo and RAGAs for evaluating LLM outputs.
Testing Mindset: Treat prompts and graphs as codeversion, diff, and test them using golden sets and regression suites.
Metrics Awareness: Maintain a strong awareness of AI evaluation metrics and how to verify RAG applications.

Security & Compliance

Red Teaming: Maintain familiarity with red teaming guardrails for AI systems.
Guardrails: Implement policy chains and guardrails using tools like OPA/Gatekeeper or Presidio for PII redaction.

Artificial Intelligence Engineer