Hey, I'm Nag 👋
Lead Engineer · InvestCloud
Los Angeles Metro
I build distributed data systems — and the LLM agents that run on top of them.
At InvestCloud I architected the Kafka/Debezium CDC pipeline (Oracle → Postgres) and an in-house ETL framework ingesting 100M+ records. Most recently I designed and built an LLM agent (Claude + Cursor + Jira + GitLab) that turns bug tickets into reviewer-assigned MRs — early results show ~40% faster resolution. 10+ years in fintech, grew with the company from startup to a $1B+ unicorn.
Targeting Senior / Staff Backend Engineering · Architect-track.
Professional Experience

  1. InvestCloud Inc.

    Lead Engineer & Product Owner

    Own the engineering roadmap for InvestCloud's data ingestion and developer-automation platforms. Architected a real-time Change Data Capture pipeline (Kafka + Debezium, Oracle → PostgreSQL) with Liquibase-driven schema evolution, and an in-house ETL framework that ingests 100M+ records across Oracle, Postgres, and SingleStore. Most recently designed and built an LLM-powered agent (Jira + Claude / Cursor + GitLab) that auto-proposes bug fixes and raises MRs — early results show ~40% faster bug resolution and 20+ engineering hours reclaimed per sprint. Containerized workloads with Docker/Kubernetes on AWS. Reviewed architecture across teams, mentored engineers, and helped scale the company from early-stage startup to $1B+ valuation with 1,000+ employees.

  2. InvestCloud Inc.

    Integration Developer

    Upgraded the core ETL framework to support multi-tenant, client-specific workflows with reusable components — improving onboarding time for new clients by 60%. Built and maintained automation for historical data loads, GitLab CI/CD pipelines, Linux server migrations, and custom client integrations.

  3. Cars.com

    Big Data Developer

    Built and maintained large-volume ETL pipelines on Apache Spark and Hadoop — improved ingestion speeds 40% and cut processing times 30%. Implemented data validation and error-handling that materially improved data quality, while enforcing integrity, confidentiality, and regulatory compliance.

Education
Oklahoma State University
M.S. in Computer Science
|
May 2015
Jawaharlal Nehru Technological University
B.Tech in Computer Science
|
May 2012
Featured Work
Production systems I've shipped, plus their open-source reference architectures.
Autonomous Bug-Fix Agent
InvestCloud
A closed-loop AI agent I designed and built at InvestCloud that converts Jira bug tickets into reviewer-assigned merge requests, end-to-end. A cron watcher polls a curated Jira project for new bugs; for each one, the agent gathers context (linked stories, repo state, recent commits, error traces). When confidence is below threshold, it comments back on the ticket asking for the specific missing context — and resumes automatically once a human responds. Once the spec is clear, it plans, implements, runs unit + integration + end-to-end tests where applicable, passes a local review-agent gate, addresses findings, re-runs tests + lint, then pushes a branch, opens a merge request, and assigns the appropriate reviewer. Early results: ~40% faster bug resolution and ~20+ engineering hours reclaimed per sprint. Stack: Claude (primary LLM) + Cursor (agentic coding), Jira API, GitLab API, internal test harnesses. Happy to walk through the architecture on a call.
Real-time Change Data Capture Pipeline
InvestCloud
Architected and shipped a real-time CDC pipeline at InvestCloud streaming Oracle → PostgreSQL with Debezium + Kafka, paired with Liquibase-driven schema evolution so application releases stay decoupled from database schema rollouts. Replaced overnight batch jobs and made downstream analytics fresh by the second. Companion in-house ETL framework supports Oracle / Postgres / SingleStore and ingests 100M+ records, monitored via real-time dashboards (Next.js, Node.js, Socket.io, Redis).
Data Drift — Config-driven ETL Framework (Java)
Open Source
An open-source, config-driven ETL framework I built in Java. Define an entire pipeline — sources (S3, Kafka, local files, RDBMS), transforms, sinks (Postgres, etc.) — in a single YAML file, no code. Extensible by design: add a new source or sink by implementing a small interface. Distilled from patterns I shipped in production at InvestCloud.
Streaming ETL Reference Architecture (Kafka + Flink + Postgres)
Open Source
End-to-end streaming ETL reference architecture I built and open-sourced: Kafka for ingest, Apache Flink (Java) for stateful real-time transforms, Postgres for the analytical sink — fully Dockerized so it stands up with one command. Mirrors the production pattern from the InvestCloud CDC pipeline without exposing proprietary code.
Skills
Languages
Java
Python
Scala
Groovy
TypeScript
JavaScript
Data & Streaming
Apache Kafka
Debezium (CDC)
Apache Spark
Apache Flink
Apache Hadoop
Apache Hive
Liquibase
Databases
PostgreSQL
Oracle
SingleStore
Snowflake
Redis
MySQL
Infrastructure & DevOps
Docker
Kubernetes
AWS
GitLab CI/CD
Nginx
Linux / Unix
AI / LLM Tooling
Claude (Anthropic)
Cursor (agentic coding)
OpenAI APIs
RAG patterns
Web & App
Next.js
React
Node.js
Socket.io
Contact
Best reached at nchat.dev@proton.me.
Preferred subject line: [Role] — [Company] — Nag. I reply within 24 hours.