Neel Shah
Tech Lead · Senior Data Engineer · AI/LLM Specialist · Researcher
Profile
Tech Lead and Senior Data Engineer with 10+ years delivering large-scale PySpark systems across two regulated, high-stakes industries — national health infrastructure and financial services credit risk. At CIHI, leads engineering of pipelines processing 1B+ Canadian health data points (registry, diagnosis, pharma) for government and NPO clients; previously at EXL / Goldman Sachs, built PySpark credit risk platforms handling Apple Card, Walmart Card, and GM Card portfolios at 1M transactions/hour, saving clients $10M+. Cloud-native across Azure, AWS, and Databricks. Now actively applying this foundation to LLM deployment and AI integration — RAG pipelines, local model deployment for privacy-first workloads, and AI-ready data curation. Dual Master's degrees (GPA 3.9/4.0), published researcher with 89+ citations, and open source creator with 1M+ downloads.
AI & LLM Expertise
- Claude API (Anthropic)
- OpenAI / GPT-4
- Prompt Engineering
- RAG pipelines
- Function Calling
- Embeddings
- LangChain
- Ollama
- LM Studio
- llama.cpp
- Mistral
- Llama 3
- Phi-3
- On-prem / PII-safe AI
- Dataset curation
- Synthetic data
- Fine-tuning prep
- RLHF data
- Annotation pipelines
- Health data for AI
- PySpark
- Apache Spark
- Databricks
- Azure
- AWS
- Large-scale ETL
- Data governance
Technical Skills
Experience
- Led the transformation of the R&D flagship product "CMG Grouper" from SAS to Python and PySpark — now processing up to 24 million records with 200+ parameters in under 60 minutes
- Engineered and deployed an end-to-end ETL pipeline for CMG Grouper, ingesting diverse data types from all Canadian hospitals at population scale (1B+ data points across registry, diagnosis, and pharma datasets)
- Serve government (federal & provincial) and NPO health clients — leading requirements gathering, client relationship management, and delivery across multiple concurrent engagements
- Lead and mentor cross-functional engineering team through full SDLC: architecture, development, code review, testing, deployment, and ongoing maintenance in an agile environment
- Define and enforce health information privacy and security standards for highly sensitive PII health data, ensuring rigorous compliance with PIPEDA and provincial health privacy legislation
- Drive project roadmaps and stakeholder communication — accountable for milestones, quality, and outcomes across the full programme
- Built and owned the core credit card policy engine using Python, PySpark, and FastAPI — handling Apple Card, Walmart Card, and GM Card fraud and credit risk portfolios
- Designed Fraud and Credit policy systems capable of processing 1 million financial transactions per hour with full PII compliance and regulatory audit trails
- Architected large-scale credit risk management REST API handling 1,000+ application requests per minute with sub-second latency
- Resolved multiple P-0 production incidents, delivering successful outcomes that saved client companies at least $10 million in combined risk exposure
- Built a Python and py-unit test automation framework that reduced end-to-end testing time by 60%
- Led technical requirements gathering, tech stack evaluation, and key engineering hiring decisions for the Goldman Sachs engagement
- Led web development team through a full waterfall-to-agile transformation, improving delivery efficiency and product quality across 42 company websites
- Designed and deployed a high-throughput microservice using Python, FastAPI, Docker, and AWS capable of handling 100,000 requests per hour
- Developed AEM virtualization of Dev and Production environments for all 42 websites using Docker, Python, and AWS — delivering $5 million USD in annual cost savings
- Built a WCAG accessibility analysis tool covering all 42 company websites using Python, REST API, and SQL, ensuring compliance and improving user experience
- Designed information architecture for KPI tracking and APIs using Python 3, REST API, and MongoDB across the full Canopy Growth digital estate
- Performed root cause analysis on multiple infrastructure incidents, improving system stability and availability to 99%+
- Built, maintained, and scaled Azure cloud infrastructure of 1,800+ servers (Windows and Linux) with 99.99% uptime SLA
- Developed automation scripts for server monitoring, patching, and maintenance using Python, REST API, and Azure
- Built real-time Power BI dashboards for Azure infrastructure monitoring, enabling data-driven operations for the platform team
- Developed and maintained Python scripts for AXIS and Moody's distributed computational environment — reducing operational cost by 5%
- Automated CI/CD pipeline using Python, Docker, and Git; reduced debugging time by 45 minutes through automated Azure environment testing
- Designed and developed a real-time airport analytical system integrating multiple hardware (LiDAR, Camera) using Python and reactive programming for operational intelligence
- Led migration of a large-scale Python 2 codebase to Python 3, modernising core airport systems without service disruption
- Transformed a legacy monolithic system into a cloud-based microservice architecture on Azure, significantly improving scalability and reliability
- Reduced backend testing time by 30 minutes by integrating automated testing into the CI pipeline
- Built a global Power BI data visualisation platform for Azure product line analytics, enabling organisation-wide data-driven decisions
- Published 3 peer-reviewed research papers on NLP, public health analytics, and distributed data systems — accumulating 89+ citations (NSERC-funded, $7,000/year Discovery Grant)
- Designed and developed Elasticsearch cluster with 20 nodes capable of searching 330 million tweets per second for real-time social media health analytics
- Built an end-to-end ETL pipeline and analytical platform on AWS, serving as a primary data source for both the Canada Health Department and internal research teams
- Developed Random Forest NLP model achieving 93.4% accuracy for population-level public health classification
- Built an asynchronous chatbot analytics API capable of handling thousands of requests per second
- Developed 5+ real-time visualisation dashboards for chatbot semantic analysis and topic extraction on AWS
- Designed a clustering algorithm for chat-based decision support systems
- Created Python and Bash tooling for call centre automation, data conversion, and API integration (REST, JSON, CRUD)
- Built real-time data analysis system for raw product cost and transportation logistics using SAP and Python
- Developed a time-series sales forecasting model for ice-cream product lines achieving 71% prediction efficiency
- Designed ETL logic and report generation pipeline for sales, cost, and inventory data in multiple formats (Excel, CSV, PDF)
- Collaborated with data warehouse leads to evaluate and redesign ETL architecture for improved performance