Job Description & Details
This is a hands-on Generative AI engineering role focused on shipping actual production-grade RAG and agentic workflows rather than just experimenting in Jupyter notebooks. If you enjoy wrangling vector databases, chaining LLMs, and figuring out how to reduce latency while keeping context accurate, this position hits the sweet spot between research and scalable software engineering.
What You'll Actually Be Doing
Your day-to-day will involve designing and deploying Retrieval-Augmented Generation architectures, building autonomous agentic systems that can reason and execute tasks, and writing robust Python code to pipe data into models. You'll spend a good chunk of your time optimizing inference pipelines, debugging why a model is hallucinating or timing out, and collaborating closely with product and data teams to turn fuzzy business requirements into deterministic AI workflows.
The Core Tech Stack
You need to know your way around Python, LangChain, and Hugging Face Transformers like the back of your hand, alongside heavy experience integrating frontier models like OpenAI GPT and Anthropic Claude. Just as importantly, you'll need deep familiarity with vector databases like Pinecone, Weaviate, or FAISS because the core challenge here isn't just calling an API; it's efficiently storing, indexing, and retrieving embeddings at scale.
Interview Expectations
Expect the hiring team to grill you on how you handle edge cases in RAG architectures, such as chunking strategy optimization and re-ranking techniques to fix retrieval noise. They'll likely ask you to debug a scenario where a custom agent loops infinitely or fails to parse tool outputs properly, testing your fundamental grasp of state management and prompt engineering under the hood.
Application Advice
Skip the generic bullet points and make sure your resume explicitly highlights production deployments involving vector search and LLM orchestration frameworks. Use keywords like RAG, LangChain, PyTorch, Pinecone, and CI/CD pipelines to ensure you clear the ATS filters, and be prepared to talk through the architectural tradeoffs you made on past machine learning projects.