Senior Data Engineer → AI Platform Engineer
Databricks Lakehouse · Azure · LLM Pipelines · Agentic RAG · Banking & Cybersecurity
I'm a Senior Data Engineer (VP-level) with 10+ years building production-grade data platforms across banking, cybersecurity, risk technology, and healthcare — and I'm now extending that foundation into AI platform engineering.
Most AI systems fail because the data layer is broken. I build both.
I've led enterprise Lakehouse programs at Tier-1 financial institutions, processing 30M+ records daily across full Bronze/Silver/Gold medallion architectures with real security tooling data. I design the metadata-driven frameworks, governance layers, and CI/CD foundations that teams actually adopt — not just POCs that get shelved.
I'm now building LLM pipelines, Agentic RAG systems, and AI-powered DataOps tools on top of the same governed data infrastructure I've spent a decade perfecting — bridging the gap between enterprise data platforms and the AI workloads companies are rushing to deploy.
ruby = {
"shipped": ["Agentic RAG Knowledge Assistant 🧠", "M&A Oracle (team capstone) 🏦"],
"currently_building": ["CyberLens 🔵", "QueryForge 🟢", "PipelineGuardian 🟡"],
"production_stack": ["Databricks", "Delta Lake", "PySpark", "Azure", "Unity Catalog"],
"ai_stack": ["Claude API", "LangGraph", "RAG", "MLflow", "ChromaDB", "RAGAS"],
"domain_expertise": ["Banking", "Cybersecurity", "Risk Technology", "Healthcare", "M&A / PE"],
"open_to": ["Senior Data Engineer", "AI Platform Engineer", "ML Platform Engineer"],
}Real systems. Real code. Running now.
| Project | What It Does | Key Stack | Status |
|---|---|---|---|
| 🧠 Claude Cert Knowledge Assistant | Domain-specific Agentic RAG system with a 3-tier adaptive execution model: Tier 0 (direct LLM), Tier 1 (single-source retrieval), Tier 2 (multi-hop decomposition across sources). Implements Corrective RAG with query rewriting, Self-RAG hallucination grading, semantic caching, and hierarchical parent-child chunking. RAGAS evaluation suite included. | LangGraph · LangChain · ChromaDB · OpenAI API · RAGAS · Brave Search · Pydantic | ✅ Live |
| 🏦 M&A Oracle (team capstone) | [Shipping Apr 20, 2026] Enterprise RAG system for private equity due diligence — surfaces contradictions between management earnings calls and SEC filing footnotes. Integrates 7+ data sources (SEC EDGAR 10-K/10-Q, USPTO patents, earnings transcripts) via a Knowledge Graph, RAG Router, and multi-step agentic reasoning. Enterprise-grade observability, audit trails, and Slack notifications. | Knowledge Graph · RAG Router · SEC EDGAR · USPTO API · LangGraph · Agentic AI · Multimodal RAG | 🔨 Shipping Apr 20, 2026 |
Bridging 10 years of enterprise Data Engineering into the AI platform layer. Each project applies production-grade patterns from real regulated environments — banking, cybersecurity, and healthcare.
| Project | What It Does | Key Stack | Status |
|---|---|---|---|
| 🔵 CyberLens | Security Data Lakehouse + Agentic RAG platform. Ingests multi-source security telemetry into a Bronze/Silver/Gold medallion architecture, then layers a LangGraph agent that answers natural language questions about enterprise security posture by reasoning over both structured Delta data and unstructured threat intel. | Databricks · Delta Lake · LangGraph · Claude API · ChromaDB · FastAPI · MLflow · Docker | 🔨 Building |
| 🟢 QueryForge | Production-grade Text-to-SQL LLMOps platform — the open-source version of what Databricks Genie and Snowflake Cortex Analyst are commercializing. Prompt versioning in MLflow, RAGAS evaluation CI/CD that fails builds on accuracy drops, SQL validation layer, and a user feedback loop for continuous improvement. | Claude API · MLflow · RAGAS · FastAPI · PostgreSQL · GitHub Actions · Streamlit · Databricks | 🔨 Building |
| 🟡 PipelineGuardian | Agentic DataOps monitor that detects pipeline anomalies (schema drift, volume drops, SLA breaches), gathers lineage context, generates LLM-powered root cause analysis via a 4-node LangGraph workflow, and auto-creates incident tickets with AI-written descriptions — turning hours of manual triage into 60-second automated resolution. | LangGraph · Claude API · Delta Lake · Docker · FastAPI · MLflow · Streamlit · Databricks | 🔨 Building |
Data Platforms & Engineering
AI & LLM Engineering
Languages
Cloud & DevOps
Databases
| 🔵 Microsoft | 🟠 Databricks | 🟤 Anthropic |
|
✅ Azure Data Engineer Associate (DP-203) ✅ Azure Data Scientist Associate (DP-100) ✅ Fabric Analytics Engineer (DP-600) ✅ Power BI Data Analyst (PL-300) ✅ Azure AI Fundamentals (AI-900) ✅ Azure Data Fundamentals (DP-900) ✅ Azure Fundamentals (AZ-900) |
✅ Lakehouse Fundamentals 🔄 Data Engineer Professional (May 2026) ✅ Building Single-Agent Apps on Databricks ✅ GenAI App Deployment & Monitoring ✅ Building Retrieval Agents on Databricks |
🔄 Claude Code Architect Foundations (Apr 2026) ✅ Claude Code 101 🔄 Building with the Claude API 🔄 Introduction to Agent Skills |
This is a live job search sprint. I'm building in public.
| 🔨 Active Build | CyberLens · QueryForge · PipelineGuardian — all three in parallel |
| 📚 Active Learning | Databricks Mosaic AI · Azure AI Foundry · RAG Architect Boot Camp (datasenseai.com) |
| 🎓 Certs in Progress | Databricks Data Engineer Professional · Claude Code Architect Foundations |
| 🌍 Available | Immediately · New Jersey / Remote / Hybrid |
| 🤝 Looking for | Senior DE · AI Platform Engineer · ML Platform Engineer |
- Databricks Mosaic AI — Vector Search, Model Serving, AI Gateway, Genie patterns
- LLMOps — Prompt versioning, RAGAS evaluation, agent observability, cost tracking
- Azure AI Foundry — Prompt Flow, Azure AI Search, multi-provider LLM abstraction
- Agentic System Design — Multi-agent orchestration, tool use, context engineering, evaluation harnesses
- RAG Architect Boot Camp — 8-week enterprise RAG solution builder (datasenseai.com)
I'm actively exploring Senior Data Engineering and AI Platform Engineering roles where I can bridge enterprise data platforms with production AI workloads.
- 🌍 Location: New Jersey — open to remote & hybrid roles across the US
- 📧 Email: roopmathi.gj@gmail.com
- 💼 LinkedIn: linkedin.com/in/roopmathi
- 🇨🇦 Work Auth: Canadian citizen · TN visa · processes at US border in 1 day · no USCIS wait · no lottery
"Most AI systems fail because of the data layer. I build both."