We are looking for a Python AI Engineer (2–3 yrs) who can build and productionize AI/GenAI solutions—especially LLM-powered applications such as RAG systems, summarization, classification, and agentic workflows. The role is engineering-led: strong Python coding, API development, deployment readiness, and basic operational practices (monitoring/evaluation/guardrails). This is not an ML platform role.

✅ Key Responsibilities

GenAI / LLM Engineering

Build LLM-powered applications (chatbots, copilots, summarization, knowledge assistants) using OpenAI/Azure OpenAI/Anthropic/Gemini or open-source LLMs.

Implement RAG pipelines: data ingestion → chunking → embeddings → vector search → prompt assembly → response generation.

Improve response quality using prompt engineering, retrieval tuning (hybrid search, metadata filters), and basic RAG evaluation practices.

ML Engineering (non-platform)

Develop and deploy ML components (classification, NLP, forecasting) using scikit-learn / PyTorch / TensorFlow as needed.

Package AI/LLM solutions into production-grade services using FastAPI/Flask.

Write clean, reusable Python modules and follow engineering best practices (testing, logging, code quality).

Deployment & Operations (LLMOps exposure)

Support deployment to cloud environments: AWS (SageMaker/ECS/Lambda) or Azure (Azure ML/AKS/App Services).

Implement basic observability: logs, error handling, latency tracking, token usage tracking (where applicable).

Assist in quality, safety, and governance practices: PII redaction, content filtering, prompt-injection mitigation, secure access controls.

Python programming (strong fundamentals, OOP, writing APIs, debugging).

Hands-on experience building GenAI/LLM solutions: RAG / embeddings / vector DB / prompt engineering.

Experience with FastAPI or Flask (building and serving APIs).

Understanding of LLM application lifecycle (prompting, evaluation, versioning, deployment basics).

Knowledge of at least one cloud platform: AWS or Azure.

Basic understanding of Git, code reviews, and deployment workflows.

Vector databases: Pinecone / Qdrant / Chroma / Weaviate / FAISS.

Frameworks: LangChain / LangGraph / LlamaIndex / Semantic Kernel.

Evaluation tools: RAGAS / TruLens / DeepEval, prompt testing frameworks.

Containerization: Docker (Kubernetes is optional).

CI/CD exposure: GitHub Actions / Azure DevOps / Jenkins.

Data pipelines: Airflow / Prefect / Databricks.

Safety tooling: Presidio, content safety filters, access control patterns.