Senior AI Engineer

We're looking for a Sr AI Engineer who has shipped AI products, not just prototyped them, with equal footing in core ML and modern GenAI. This role sits at the intersection of engineering rigor and product ownership, you'll build, deploy, and operate ML and LLM-powered applications on Azure, with real accountability for what happens after go-live: accuracy, cost, latency, safety, drift, and uptime.

If you've only worked in notebooks or built demos that never saw production traffic, this isn't the role. If you've had to explain to a stakeholder why a model started hallucinating in week three or had to design a rollback plan for a prompt change, we want to talk to you.

What You'll Do

• Own end-to-end deployment of AI applications on Azure — from classical ML models and Azure OpenAI Service integrations through to production release, monitoring, and iteration.
• Build and evaluate core ML models where GenAI isn't the right tool — classification, regression, forecasting, clustering, or recommendation problems using traditional ML  techniques.
• Design and implement guardrails — content filtering, prompt injection defence, PII redaction, output validation, and human-in-the-loop checkpoints for high-risk actions.
• Build and maintain MLOps/LLMOps pipelines — CI/CD for models and prompts, feature engineering and data pipelines, automated evaluation harnesses, versioning for models/prompts/embeddings/fine-tunes, and rollback mechanisms.
• Manage the model lifecycle — model selection and routing (classical ML vs. smaller LLMs vs. frontier models by task complexity and cost), performance benchmarking, cost-per-call/cost-per-inference tracking, and deprecation/upgrade planning.
• Implement observability — logging, tracing, and alerting for LLM applications (token usage, latency, hallucination/error rates, user feedback loops).
• Architect RAG and agentic systems — vector store design, retrieval tuning, orchestration frameworks (LangGraph, Semantic Kernel, or equivalent), and multi-agent workflows where applicable.
• Collaborate cross-functionally with product owners, architects, and business stakeholders to translate requirements into scoped, deployable AI features.
• Contribute to AI governance — support responsible AI reviews, model risk assessments, and documentation required for enterprise sign-off.

What You Bring

Must-Have

• Minimum of 3–4 years of hands-on software/ML engineering experience, with at least 2 years specifically on GenAI/LLM applications taken to production.
• Solid grounding in core AI/ML fundamentals — supervised/unsupervised learning, model evaluation metrics, feature engineering, handling class imbalance/overfitting, and knowing when a classical ML model beats an LLM for the job.
• Hands-on experience with standard ML libraries (scikit-learn, XGBoost / LightGBM, pandas, NumPy) and at least one deep learning framework (PyTorch or TensorFlow).
• Strong working knowledge of the Azure AI/ML stack: Azure Machine Learning, Azure OpenAI Service, Azure AI Foundry, Azure AI Search, and Azure App Service/Functions for deployment.
• Practical experience with MLOps/LLMOps tooling — CI/CD pipelines, containerization (Docker), model/prompt versioning, experiment tracking, and automated testing/evaluation frameworks.
• Demonstrated experience building guardrails and safety layers in production — not just theoretical familiarity (e.g., Azure AI Content Safety, custom validation layers, jailbreak/prompt-injection mitigation).
• Solid Python engineering skills — clean, testable, production-grade code, not notebook scripts.
• Experience with at least one orchestration framework: LangGraph, Semantic Kernel, LangChain, or similar.
• Understanding of RAG architecture — chunking strategies, embedding models, vector databases, retrieval evaluation.
• Comfort with monitoring/observability tooling (Application Insights, or equivalent) for live AI systems, including model performance monitoring and drift detection.

Good to Have

• Exposure to Copilot Studio or Power Platform for low-code AI extensions.
• Experience with voice-based or multimodal AI applications.
• Familiarity with enterprise AI governance frameworks and responsible AI principles.
• Prior experience in AEC, GCC, or large enterprise delivery environments.
• Contributions to internal upskilling, documentation, or mentoring within an AI team.

What Sets Strong Candidates Apart

We're specifically screening for deployment maturity across both classical ML and GenAI, over research depth alone. In interviews, be ready to talk through:

• A time you had to redesign a guardrail after it failed in production.
• How you've tracked and controlled LLM cost at scale (tiered model routing, caching, batching).
• Your approach to versioning and rolling back a prompt or model change without breaking downstream consumers.
• How you've measured and reduced hallucination or drift in a live system.
• A time you chose (or should have chosen) a classical ML model over an LLM, and why.

BGV:

• Employment with WSP India is subject to the successful completion of a background verification (“BGV”) check conducted by a third-party agency appointed by WSP India.

• Candidates are advised to ensure that all information provided during the recruitment process — including documents uploaded — is accurate and complete, both to WSP India and its BGV partner”.