AIML Engineer with GCP

Experience: 7+ years Preferred Locations: Chennai and Coimbatore Strong hands-on experience in AI/ML and Generative AI technologies Job descriptio n Build LLM agent systems with tool use, memory, and multi-step orchestration - handle retries, timeouts, and graceful degradati onImplement full RAG pipelines: PDF/document/SCORM/Audio/Video file parsing , chunking strategies, embedding generation (OpenAI, or Vertex AI embeddings), and vector stora geBuild retrieval optimization layers either with hybrid search, cross-encoder re-rankin g.Develop file ingestion and transfer pipelines using GCP Cloud Storage, Pub/Sub triggers, and Cloud Functions or Cloud Run for async processing (Good to have - GCP or Azur e)Productionize pipelines with latency SLOs, token cost tracking, output validation (Guardrails AI, custom classifiers), and model versionin g.Build eval frameworks using RAGAS, custom LLM-as-judge pipelines, or human feedback loops to measure retrieval quality, faithfulness, and answer relevan ceInstrument AI pipeline observability trace LLM calls, log retrieved chunks, monitor embedding drift, and alert on quality regressi on Experience: 7+ years Preferred Locations: Chennai and Coimbatore Strong hands-on experience in AI/ML and Generative AI technologies Job descriptio n Build LLM agent systems with tool use, memory, and multi-step orchestration - handle retries, timeouts, and graceful degradati onImplement full RAG pipelines: PDF/document/SCORM/Audio/Video file parsing , chunking strategies, embedding generation (OpenAI, or Vertex AI embeddings), and vector stora geBuild retrieval optimization layers either with hybrid search, cross-encoder re-rankin g.Develop file ingestion and transfer pipelines using GCP Cloud Storage, Pub/Sub triggers, and Cloud Functions or Cloud Run for async processing (Good to have - GCP or Azur e)Productionize pipelines with latency SLOs, token cost tracking, output validation (Guardrails AI, custom classifiers), and model versionin g.Build eval frameworks using RAGAS, custom LLM-as-judge pipelines, or human feedback loops to measure retrieval quality, faithfulness, and answer relevan ceInstrument AI pipeline observability trace LLM calls, log retrieved chunks, monitor embedding drift, and alert on quality regressi on