01Overview
Discover your future at Citi Working at Citi is far more than just a job. A career with us means joining a team of more than 230,000 dedicated people from around the globe. At Citi, youll have the opportunity to grow your career, give back to your community and make a real impact. Job Overview Minimum QualificationsBachelor's degree in Computer Science, Data Science, Artificial Intelligence, or a related quantitative field.812 years of experience as a Data Scientist or equivalent role, with at least 2 years of specialized, hands-on experience in Generative AI, including leading technical development and mentoring teams.Demonstrable experience across the full lifecycle of production-level GenAI projects from ideation and prototyping through deployment, monitoring, and ongoing maintenance in live environments. Proof-of-concept work alone is insufficient.Core ResponsibilitiesWorking with financial and enterprise data, applying modern NLP and GenAI techniques to solve business problems.Designing, refining, and systematizing prompt engineering strategies for large language models (LLMs), including structured prompting, chain-of-thought, and few-shot/zero-shot approaches.Collaborating with business stakeholders to translate requirements into GenAI-powered solutions.Developing, testing, and maintaining production-grade Python code for GenAI applications.Integrating with vector databases (e.g., Pinecone, Weaviate, Milvus, pgvector, Qdrant) for retrieval-augmented generation (RAG) pipelines.Building, monitoring, and optimizing MLOps/LLMOps pipelines for continuous model deployment and observability.Researching and evaluating emerging GenAI technologies, frameworks, and best practices to maintain competitive advantage.Troubleshooting and debugging GenAI models and agentic systems in production, including rapid identification and resolution of issues in real-world deployments.Communicating complex AI/ML concepts clearly to non-technical stakeholders, translating technical jargon into actionable business terms.Participating in and leading team meetings, design reviews, and architecture discussions.Technical SkillsProgramming & FoundationsExpert-level Python proficiency, including:Core Python: data structures (lists, dictionaries, sets), algorithms, object-oriented programming, async programming, file handling, and exception handling.Scientific computing: NumPy, Pandas, SciPy.Machine LearningScikit-learn, XGBoost, LightGBM.Strong understanding of advanced modeling techniques, model evaluation, hyperparameter tuning, and deployment strategies.Deep LearningPyTorch (preferred/primary), TensorFlow/Keras.Familiarity with training, fine-tuning, and inference optimization for neural network architectures.Generative AI (Updated for Current Landscape)AreaKey Technologies & ConceptsLLM Frameworks:Hugging Face Transformers, LangChain, LlamaIndex, Semantic KernelAgentic AI:LangGraph, CrewAI, AutoGen, tool-use/function-calling patterns, multi-agent orchestrationLLM Architectures:Transformer architectures (decoder-only, encoder-decoder), Mixture-of-Experts (MoE), multimodal models (vision-language models)RAG (Retrieval-Augmented Generation):Advanced RAG patterns (hybrid search, re-ranking, query decomposition, contextual retrieval), chunking strategies, embedding models (e.g., OpenAI, Cohere, open-source sentence-transformers)Vector DatabasesPinecone, Weaviate, Milvus, Qdrant, pgvector, ChromaDBPrompt Engineering:Structured prompting, chain-of-thought, ReAct, few-shot/zero-shot, prompt chaining, guardrails and output parsingModel Serving & OptimizationvLLM, TGI (Text Generation Inference), ONNX Runtime, quantization (GPTQ, AWQ, GGUF), model distillationEvaluation & ObservabilityLLM evaluation frameworks (RAGAS, DeepEval, custom evals), LLM observability tools (LangSmith, Arize Phoenix, Weights & Biases), red-teaming and safety testingAPI DevelopmentFastAPI, RESTful and streaming API design for GenAI applications, WebSocket integrationResponsible AIBias detection and mitigation, content safety filters, hallucination reduction techniques, AI governance frameworksMLOps / LLMOpsCI/CD for ML/GenAI pipelines (e.g., GitHub Actions, GitLab CI).Experiment tracking and model registry (MLflow, Weights & Biases).Containerization and orchestration: Docker, Kubernetes.Infrastructure-as-code and deployment automation.Cloud PlatformsProficiency in at least one major cloud platform's AI/ML services:AWS (Bedrock, SageMaker, Lambda)Azure (Azure OpenAI Service, Azure AI Studio, Azure ML)GCP (Vertex AI, Gemini API)Soft SkillsExcellent communication and collaboration skills both written and verbal with the ability to effectively convey technical concepts to diverse audiences, including senior leadership and business partners.Ability to articulate the challenges, trade-offs, and successes of deploying GenAI solutions at scale.Proactive approach to continuous learning in the rapidly evolving GenAI landscape.Preferred QualificationsMaster's or Ph.D .
