Role: Data Scientist-Agentic AI / GenAI Engineer
Experience : 6 years in AI/ML/data science/software engineering, with 3 years in GenAI, LLM, RAG, conversational AI, or ML productionisation.
Common Job Description :
Strong Python.
API development using FastAPI, Flask, or similar.
Understanding of LLMs, embeddings, vector search, prompt design, evaluation, and hallucination control.
RAG architecture: ingestion, chunking, embeddings, retrieval, ranking, grounding, citations, evaluation.
MLOps / LLMOps basics: model deployment, monitoring, evaluation, versioning, observability.
Security and governance basics: IAM, PII handling, prompt injection risks, data leakage, approval workflows.
Ability to build real working prototypes and production-ready services.
Short JD : Agentic AI / GenAI Engineers who can design and deploy secure, production-grade AI agents using Google Cloud AI stack or equivalent GenAI frameworks.
Notes : GenAI/Python/RAG profiles MUST and grooming possible on ADK/Vertex/Gemini Enterprise
Alternatively, can try for :
Python backend engineers with solid LLM/RAG project experience.
ML engineers with Vertex AI and production deployment experience.
Strong LangChain/LlamaIndex engineers who can ramp up on ADK.
Detailed JD :
Generic Skills (Must Have)
Python, FastAPI, REST APIs, async processing.
LLM application development
RAG implementation with vector databases
Prompt engineering, tool calling, function calling, structured outputs.
LLM security: prompt injection, data leakage, access control, guardrails.
GCP Skills (Must Have)
VertexAI : Alternative vector databases: Vector Search, Pinecone, Weaviate, FAISS, Chroma, pgvector, or equivalent.
gemini
Agent Orchestration using ADK: Alternatives: LangChain, LlamaIndex, Semantic Kernel, CrewAI, AutoGen, or equivalent.
Cloud Run: Production deployment on Cloud Run, GKE, or equivalent.
Evaluation using Vertex AI : Alternatives: Evals, RAGAS, custom eval frameworks, golden datasets, regression tests.
Nice to have (Trainable)
Google Agent Development Kit.
Agent Engine / Gemini Enterprise Agent Platform.
Model Armor.
Agent observability and tracing.
Multi-agent architecture.
Human-in-the-loop approval flows.
Enterprise knowledge graph / search integration.
