What Skills to Look For in a Generative AI Engineer

Core Skills (Non-Negotiable)

• ✦ Python (transformers, langchain, llamaindex, huggingface)
• ✦ Understanding of transformer architecture (attention, embeddings, context windows)
• ✦ Experience with LLM APIs (OpenAI, Anthropic, Groq, together.ai)
• ✦ Prompt engineering (system prompts, few-shot, chain-of-thought, react)

RAG & Retrieval Skills

Must have:

• ✦ Vector database experience (Pinecone, Weaviate, Qdrant, Chroma, PGVector)
• ✦ Embedding models (text-embedding-3-small, voyage, bge, e5)
• ✦ Chunking strategies (semantic, recursive, document-aware)
• ✦ Multi-vector retrieval, hybrid search, re-ranking

Fine-Tuning & Model Optimization

For production systems:

• ✦ Fine-tuning open-source models (Llama, Mistral, Qwen)
• ✦ Parameter-efficient fine-tuning (LoRA, QLoRA)
• ✦ Dataset preparation and curation
• ✦ Model evaluation before/after fine-tuning

Evaluation & Testing

Critical for production:

• ✦ LLM-as-judge evaluation frameworks
• ✦ Metrics: accuracy, relevance, hallucination rate, faithfulness, answer relevancy
• ✦ A/B testing for prompt and model changes
• ✦ Human evaluation workflows

Production & Deployment

Senior candidates have:

• ✦ Inference optimization (vLLM, TensorRT, quantization)
• ✦ Cost monitoring and optimization (token usage, caching)
• ✦ Latency optimization for real-time applications
• ✦ Monitoring (hallucination detection, drift, performance)

Portfolio Signals That Matter

• ✦ Production RAG applications with real users
• ✦ Open-source contributions to LLM tooling
• ✦ Published technical write-ups on GenAI systems
• ✦ Evaluation frameworks and benchmark reports
• ✦ Evidence of cost and latency optimization work

Hiring Red Flags

• ✦ Only prompt engineering experience
• ✦ No production deployment examples
• ✦ Unable to explain retrieval evaluation metrics
• ✦ No understanding of token cost management
• ✦ Cannot discuss hallucination mitigation strategies

Hire the Complete Package