1 day ago
Pune, IndiaStaff+
Responsibilities
- Architect, build, and deploy production generative AI systems using RAG, autonomous agents, semantic search, multimodal workflows, LLMs, prompt engineering, fine-tuning, evaluation benchmarks, and guardrails.
- Lead predictive ML architecture for anomaly detection, graph neural networks, supervised and unsupervised classification, fraud detection, and real-time risk assessment.
- Design resilient, low-latency, horizontally scalable Python microservices and APIs for high-throughput transactional traffic.
- Integrate vector databases, distributed caching, event streaming, and distributed processing technologies.
- Implement MLOps and LLMOps practices including model lifecycle orchestration, automated retraining, inference monitoring, observability, and model drift monitoring.
- Develop AI solutions for identity theft, account takeover, synthetic identity fraud, card-not-present fraud, transaction anomalies, fraud triage, and investigative summarization.
- Ensure adherence to FCRA, GLBA, SR 11-7, Fair Lending, and explainable AI requirements.
- Mentor senior- and mid-level data scientists and software engineers and drive technical roadmaps, proofs of concept, framework evaluations, and architectural proposals.
Requirements
- Bachelor’s or Master’s degree in Computer Science, Data Science, Artificial Intelligence, Applied Mathematics, or a related quantitative field.
- 10 to 12 years of professional experience in software engineering and production AI/ML system development.
- Advanced Python proficiency, including multiprocessing, asyncio, design patterns, profiling, packaging, and clean code practices.
- Deep practical experience with PyTorch, TensorFlow, Scikit-learn, XGBoost, and LightGBM.
- Demonstrated success building production generative AI applications with frameworks such as LangChain, LlamaIndex, CrewAI, and AutoGen.
- Experience with embedding models, dense retrieval, rerankers, vector indexing, LLM API integration, function calling, context caching, and prompt optimization.
- Strong hands-on experience with AWS, GCP, or Azure; Docker and Kubernetes; and distributed processing with PySpark or Ray.
- Proven ability to orchestrate end-to-end ML lifecycles from data validation through real-time inference monitoring.
- Preferred experience with fraud rule engines, anomaly detection platforms, predictive risk models, transaction monitoring, biometric behavioral analytics, and fraud consortium datasets.
- Preferred experience with Google Gemini and Anthropic Claude APIs and with model risk management, explainable AI, and PII masking.
Benefits
- Full-time position based in the company’s Chennai center with a hybrid work model.
Tech Stack
Apache KafkaApache SparkBashDockerElasticsearchFastAPIFlaskGitHub ActionsGrafanagRPCHugging Face TransformersKubernetesLightGBMMLflowPandasPrometheusPythonPyTorchRabbitMQscikit-learnSQLTensorFlowTerraformXGBoost
Categories
About Citi
Citi is a public financial-services company offering consumer and institutional banking, credit cards, wealth management, treasury and trade solutions, and capital-markets services. It serves individuals, corporations, financial institutions, and governments in more than 160 countries and jurisdictions, earning interest and fee income from lending, payments, trading, and advisory. Founded in 1812 and headquartered in New York, it trades on the NYSE under the ticker C.