Senior AI Solutions Engineer

Required Qualifications

Core Programming & Tooling

·        Strong programming proficiency in Python, producing production-quality, well-structured code.

·        Proficient in SQL; experience with large-volume datasets in Spark and/or Snowflake.

·        Comfortable scripting in Python and Bash for automation and tooling.

·        Expertise with Git for version control, branching strategies, code review, and pull request workflows.

·        Familiarity with experiment tracking and model versioning (e.g., MLflow).

Snowflake

·        Data lake vs data warehouse architecture (medallion architecture: bronze/silver/gold)

·        API integration patterns for ingesting from source systems

·        Schema evolution and handling semi-structured data

·        Advanced SQL (window functions, CTEs, recursive queries, semi-structured data with VARIANT/JSON)

·        Warehouse sizing, scaling, and cost optimization

·        Data sharing, Time Travel, Zero-Copy Cloning

·        Snowpipe for streaming/continuous ingestion

·        Role-based access control (RBAC), security, and data governance

·        Query performance tuning (clustering keys, micro-partitions, query profiling)

·        Streams and Tasks for CDC and orchestration

·        Snowpark (Python/Java integration) is increasingly valuable

·        Model development (staging/intermediate/mart layers)

·        Jinja templating, macros, and incremental models

·        Snapshots for SCD Type 2

·        dbt testing (schema tests, dbt-utils)

·        Version control: Git/GitHub or Azure Repos, branching strategies

·        Orchestration: Airflow, Azure Data Factory, or Dagster

·        Data modeling: Dimensional modeling (Kimball), star/snowflake schemas

·        CI/CD: Automated testing and deployment for dbt models and pipelines

Machine Learning & Statistical Modeling

·        Hands-on experience building and deploying ML on large-scale production datasets.

·        Regression, classification, clustering, and ensemble methods.

·        Time series forecasting.

·        Optimization modeling (linear programming, mixed-integer programming).

·        Feature engineering, data wrangling, backtesting, and statistical analysis on structured data.

·        Sound evaluation practice: proper train/validation/test splits, leakage avoidance, and forward/out-of-sample validation.

LLM Applications & AI Agents

·        AI agent development: building agentic frameworks using Azure AI Foundry and Azure model endpoints, including tool-use patterns and multi-step orchestration.

·        Experience with orchestration frameworks such as LangChain or LlamaIndex.

·        Prompt engineering and chain-of-thought design for complex, multi-step workflows.

·        RAG pipeline development: chunking, vector embedding/indexing (Azure AI Search), and response generation using LLM endpoints.

·        Integrating LLMs and agents with structured data sources (SQL/Snowflake), tools, and external systems/APIs.

·        Grounding, citation, and evaluation techniques to control hallucination and verify model outputs.

Document Processing, OCR & NLP

·        OCR and document parsing using Azure Document Intelligence / Form Recognizer (or equivalent).

·        Extracting and structuring information from unstructured and semi-structured documents at scale (PDFs, scans, forms, tables).

·        Text processing and NLP fundamentals: entity extraction, classification, summarization, and semantic search.

·        LLM-based extraction, comparison, and term/clause analysis across long documents.

·        Handling document layout, multi-page context, and noisy/low-quality inputs.

Azure, Containers & Compute

·        Core Azure resources: Azure Machine Learning, Azure AI Studio, Azure OpenAI Service, and Azure storage/compute.

·        Docker: building, optimizing, and managing container images for ML and AI applications.

·        Azure Container Apps (ACA): containerized deployment, KEDA-based autoscaling, and replica management.

·        Azure Container Registry (ACR): image management, automated builds, and vulnerability scanning.

·        Familiarity with serverless compute such as Azure Functions for targeted tasks.

MLOps, CI/CD & Deployment

·        Hands-on DevOps/MLOps experience in production environments.

·        CI/CD pipeline design and maintenance for ML and LLM workloads using GitHub Actions or Azure DevOps.

·        Infrastructure as Code with Terraform or Bicep for reproducible Azure environments.

·        Managing environment promotion and configuration across dev, staging, and prod.

·        Azure Machine Learning model lifecycle: model registry, automated retraining pipelines, model versioning, and endpoint management.

·        Deployment strategies (blue-green, canary), rollback procedures, and automated testing for ML pipelines (integration, load, security).

·        Deploying and scaling LLM-based applications and agent infrastructure on Azure, with attention to latency, throughput, and cost.

Monitoring, Observability & Delivery

·        Delivering production-ready models and services with proper testing, validation, and monitoring.

·        Azure Monitor and Application Insights: logging, tracing, custom metrics, and alerting.

·        Model performance monitoring: data drift detection, prediction quality tracking, and automated alerts.

·        Basic UI skills (Streamlit or Gradio) for quick prototyping and stakeholder demos.

Soft/Leadership Skills (Senior-level expectations )

• Technical roadmap ownership / architecture decisions

• Setting engineering standards (code review bar, testing standards) across both DE and DS work

• Ability to review code/PRs and mentor junior engineers

• Stakeholder communication (translating technical concepts to business users)

• Documentation and knowledge-sharing practices

• Cost optimization mindset (especially critical in Snowflake/Azure environments)

• Cross-functional collaboration with analytics, BI, and data science teams

• Incident management and troubleshooting production data issues