Data Scientist/Architect - Python

<p><p><b>Architect & Build Full-Stack AI Solutions :</b><br/><br/>- Design and build end-to-end machine learning systemsincluding pipelines, feature stores, training/inference services, and monitoring.<br/><br/>- Architect scalable, modular applications that embed foundation models, retrieval-augmented generation (RAG), and agentic workflows directly into product experiencesall on Google Cloud Platform (GCP) or Oracle Cloud Infrastructure (OCI).<br/><br/>- Develop privacy-preserving ML systems that align with food industry compliance and traceability standards.<br/><br/>- Build explainable AI frameworks to enable business stakeholders to trust and validate model outputs.<br/><br/>- Design AI architectures that deliver tenant-aware intelligence while ensuring strict data isolation and performance SLAs across our multi-tenant SaaS platform.<br/><br/></p><p><b>Hands-On Engineering :</b><br/><br/>- Develop robust data pipelines using BigQuery, Dataflow, and Apache Beam.<br/><br/>- Train, tune, and deploy ML models using TensorFlow, PyTorch, and Scikit-learn.<br/><br/>- Build full-stack AI-native applications with Python, Java, React/Angular, and GCP-native services.<br/><br/>- Implement robust MLOps pipelines that ensure reproducible training, CI/CD deployment, cost-efficient scaling, and real-time model monitoring.<br/><br/><b>Requirement :</b><br/><br/>- 10+ years in software/ML engineering; 5+ years leading architecture and engineering teams.<br/><br/>- Proven experience building production-ready AI/ML pipelines and applications end-to-end.<br/><br/>- Expertise in GenAI frameworks (OpenAI, Vertex AI), multi-agent architectures, and MLOps tooling.<br/><br/>- Experience operationalizing LLMs for enterprise use cases: hallucination reduction, security, cost optimization, and monitoring.<br/><br/>- Track record of shipping high-quality intelligent products quickly and at scale.<br/><br/>- Strong team leadership, communication, and cross-functional collaboration skills.<br/></p><br/></p> (ref:hirist.tech)