About this role
WSP seeks a Senior AI Engineer in Bengaluru or Noida, India, under its Hybrid Work Model. The role combines classical machine learning and generative AI engineering with ownership of production applications on Azure. The engineer is accountable for accuracy, cost, latency, safety, drift and uptime after launch; experience limited to notebooks or prototypes is insufficient.
Responsibilities
- Build, deploy, monitor and improve ML and LLM-powered applications, including Azure OpenAI Service integrations. Develop and evaluate classical models for classification, regression, forecasting, clustering and recommendations when they are more suitable than GenAI.
- Design production guardrails for content filtering, prompt-injection defense, PII redaction, output validation and human review of high-risk actions. Build CI/CD, feature and data pipelines, automated evaluation, versioning and rollback for models, prompts, embeddings and fine-tunes.
- Select and route models by task complexity and cost; benchmark performance and track inference costs. Implement logging, tracing and alerts for token use, latency, hallucinations, errors and user feedback. Architect RAG and, where appropriate, agentic systems using retrieval tuning, vector stores and orchestration frameworks.
- Work with product owners, architects and business stakeholders to deliver scoped AI features. Support responsible AI reviews, model-risk assessments and enterprise sign-off documentation.
Required qualifications
A minimum of 3–4 years of hands-on software or ML engineering experience, including at least 2 years taking GenAI/LLM applications into production. Strong Python engineering, core supervised and unsupervised ML, evaluation, feature engineering and production safety experience are required. Candidates need experience with standard ML libraries; PyTorch or TensorFlow; Azure's AI/ML and deployment services; MLOps/LLMOps practices including Docker, testing and experiment tracking; an orchestration framework such as LangGraph, Semantic Kernel or LangChain; RAG architecture; and live-system observability and drift detection.
Preferred
Copilot Studio or Power Platform, voice or multimodal AI, enterprise AI governance, AEC or GCC or large-enterprise delivery, and AI-team mentoring or documentation. Employment is subject to successful third-party background verification.