Sr AI Engineer
CitiusTechAbout the role
Responsibilities
Design, build, and productionize ML models for personalization, forecasting, anomaly detection, and NLP on GCP (Vertex AI, BigQuery ML, Dataflow).
Develop scalable data and feature pipelines; implement feature stores (Feast) and streaming ingestion (Pub/Sub, Kafka) for real-time inference.
Implement agentic AI patterns: multi-agent orchestration, tool-use, retrieval-augmented generation (RAG), and function calling with guardrails (policy/PII filters).
Optimize online inference for latency and cost using Vertex AI Endpoints, GPU/TPU where applicable, and autoscaling on GKE. Establish experiment frameworks (A/B, interleaving, bandits) and offline evaluation (precision/recall, ROC-AUC, NDCG, MAP). Integrate vector search (Vertex Matching Engine / FAISS / Elasticsearch) for semantic retrieval and recommendations.
Ensure privacy, security, and compliance (GDPR/CCPA); apply differential privacy where needed and follow model governance practices.
Document designs, review code, and collaborate with product, data engineering, and platform teams.
Desired Technical Skills Languages:
Python (primary), Java/Scala (nice to have).
Frameworks: TensorFlow, PyTorch, scikit-learn; JAX optional.
Pipelines & Features: Apache Beam/Dataflow, Airflow/Cloud Composer, Feast.
RAG & Agents: LangChain/LlamaIndex, function calling, toolformer patterns, vector DBs (FAISS, Chroma, Elasticsearch).
Data & Streaming: BigQuery, Spark, Kafka, Pub/Sub.
Serving: Vertex AI Endpoints, KFServing, Triton; Docker, Kubernetes (GKE).
Observability: MLflow, Vertex Experiments/Model Registry, Prometheus, Grafana.
Cloud: GCP (Vertex AI, BigQuery, GCS) primary; familiarity with AWS (SageMaker) and Azure ML.
Preferred Experience & Capabilities
Delivered ML systems at scale (batch + real-time) with measurable business impact.
Hands-on with vector search, RAG, and agentic workflows using tools/actions and governance (safety filters, jailbreak protection).
Expertise in feature engineering, data quality, and drift detection (data/model).
Strong understanding of IR/ranking metrics and online experimentation.
Ability to mentor, perform code reviews, and contribute to architectural decisions.
Primary development environment is GCP (Google Cloud Platform).
Flexibility to work with AWS and Azure is desired for portability and interoperability.
Responsibilities
- Design, build, and productionize ML models for personalization, forecasting, anomaly detection, and NLP on GCP.
- Develop scalable data and feature pipelines; implement feature stores and streaming ingestion for real-time inference.
- Implement agentic AI patterns and optimize online inference for latency and cost.
- Establish experiment frameworks and integrate vector search for semantic retrieval and recommendations.
- Ensure privacy, security, and compliance; apply differential privacy where needed.
- Document designs, review code, and collaborate with product, data engineering, and platform teams.
Qualifications
- Proficient in Python; knowledge of Java/Scala is a plus.
- Experience with TensorFlow, PyTorch, and scikit-learn.
- Familiarity with Apache Beam/Dataflow, Airflow, and Feast.
- Hands-on experience with vector search and agentic workflows.
- Strong understanding of IR/ranking metrics and online experimentation.
Skills mentioned
About CitiusTech
CitiusTech is a global technology services, consulting, and business solutions enterprise 100% focused on the healthcare and life sciences industry. We enable 140+ enterprises to build a human-first ecosystem that is efficient, effective, and equitable. Leveraging deep domain expertise and next-generation technologies including AI, Cloud, Data, and Intelligent Automation, we assist our clients to realize their vision, accelerate transformation, and achieve business outcomes. With 7,700+ healthcare technology professionals worldwide, CitiusTech powers digital innovation, business transformation, and industry-wide convergence through next-generation technologies, solutions, and products.