Resume
James Ding
Data Platform & Applied AI Engineer
Summary
I design and build production data platforms and applied-AI systems end-to-end. At AirSprint I stood up the company's data warehouse and the entire infrastructure beneath it — ingestion server, database, networking, deploy pipeline, and monitoring — plus its internal AI assistant and API layer, solo. Deep private-aviation domain knowledge paired with disciplined engineering: design → build → deploy → verify. Go, Python, TypeScript, AWS.
Work Experience
- Designed and built the company's data warehouse from scratch — a medallion-architecture (Bronze/Silver/Gold) platform on PostgreSQL/RDS consolidating Salesforce, FL3XX flight-ops, and HelpScout (~472K records across 22 objects, ~142K flight legs) — to retire a fragile Snowflake/dbt pipeline, with key metrics reconciled exactly to legacy numbers
- Provisioned and operate the full platform infrastructure solo: in-VPC EC2 ingestion server, S3 data lake with lifecycle retention, VPC/security-group networking with private database paths, a key-less security model (IAM instance roles + Secrets Manager, zero static credentials), a cross-compile → S3 → SSM deploy pipeline, and SNS/SES failure alerting
- Built a generic CDC/ELT ingestion pipeline in Go — 15-minute incremental upserts on modstamp watermarks into a self-seeding jsonb landing table, with daily versioned-CSV exports and curated Silver transforms — replacing ~5 legacy full-snapshot jobs with one uniform mechanism
- Designed and built Jetty, the internal AI agent platform serving ~400 employees at flat infrastructure cost (zero per-seat licensing): one agent with a governed tool belt — document RAG with hybrid retrieval (pgvector + full-text, RRF fusion, Cohere rerank), text-to-SQL over the warehouse, an email-gated read-only Salesforce SOQL tool, and an employee directory — on AWS Lambda (Docker) + Bedrock; evolved from a router + specialist-agents design, validated by a golden-question regression suite and stress tests
- Built the company's OpenAPI-driven Go API server (scoped API-key auth) as the single app data path, decoupling serving from ingestion; sole developer of the internal ops platform (Go + Next.js) — shipped passport MRZ scanning with eAPIS conversion (AWS Textract), employee empty-leg booking, a per-diem/empty-leg payroll engine, bitmask route-level RBAC including a ~1,600-user owner-portal roster, and fleet-utilization and crewing dashboards
- Absorbed an external Salesforce contractor's work into the internal platform, replicating the Salesforce workflows teams relied on in-app and sharply cutting Salesforce license spend; leading a contract-preserving strangler rewrite to bring the contractor-built client booking application in-house
- Shipped live integrations and automation end-to-end: SageMaker Prophet revenue forecasting, Graph webhook-driven SharePoint → S3 mirroring, Salesforce ↔ platform JWT sync, and automated vendor-contract management for Legal
- Built internal Streamlit applications automating HR and Accounting data workflows, reducing manual data preparation and standardizing report generation
- Designed an event-driven pipeline (Power Automate, AWS Lambda, S3) automating SharePoint → Salesforce ingestion, eliminating manual accounting entry; supported early data-warehouse architecture planning
Education
University of Calgary
Calgary, AB
British Columbia Institute of Technology
Burnaby, BC
University of Illinois Urbana-Champaign
Champaign, IL
Technical Skills
AI / LLM Engineering: RAG (hybrid pgvector retrieval, RRF fusion, rerank), Agent & tool-use design, Text-to-SQL, LLM router + specialist architectures, Evals / golden-set regression, Prompt design, Embedding pipelines, AWS Bedrock
Languages: Python, Go, TypeScript/React, SQL (Postgres, SOQL)
Data Platform: Data-warehouse design, Medallion architecture, CDC/ELT ingestion, PostgreSQL (AWS RDS), pgvector, S3 data lake, Data-quality & reconciliation
Cloud & Infra (AWS): Lambda, Bedrock, S3, RDS, EC2, VPC & IAM (key-less, Secrets Manager), SSM deploy pipelines, API Gateway, Textract, SNS/SES alerting
Enterprise Integrations: Salesforce (SOQL, JWT), FL3XX aviation-ops API, Microsoft Graph / Azure AD SSO