Engineering
MAINFRAME × AI × IMMERSIVE WEB
RashmiRanjan Dash
Mainframe Technology Lead · AI Solutions Engineer · Creative Technologist
01 · THE ORIGIN CODE
Every systemcarries a genome.
Twelve and a half years spent reading the DNA of enterprise systems — healthcare claims engines, payment platforms, fund management and trade settlement cores — and rewriting it for the AI era. From z/OS bedrock to AWS and retrieval-augmented intelligence, the thread is the same: understand deeply, modernize fearlessly, build beautifully.
02 · THE ENTERPRISE BACKBONE
A city of systems,kept alive & reborn.
Technology Lead
End-to-end mainframe development and modernization for product and payment systems at a leading financial services client. Leads the Mainframe Knowledge Continuity program, assessing 150+ modules and generating architecture docs and Java modernization stories with AWS, Claude AI and RAG. Tuned DB2 stored procedures to cut batch runtime costs by up to 90%, led recovery of a $120M+ healthcare production defect, and automated vendor job setup with REXX.
Technology Analyst · Mainframe Developer
Built and supported mainframe applications for healthcare claim processing, fund management and trade settlement. Resolved 500+ incidents at 98%+ SLA compliance, reverse-engineered 50K+ lines of COBOL for knowledge transfer and migration planning, and ran DEV → UAT → PROD release cycles with zero-downtime deployments.
02½ · THE MODERNIZATION BENCH
From COBOL decksto modern systems.
Every source member runs the same line: an LLM parses it, a RAG index files it, its calls, files and jobs join the dependency map, and the generator writes the outputs an Enterprise Analyzer or CAST run would give — by AI.
03 · THE INTELLIGENCE LAYER
Teaching legacysystems to think.
RAG & Knowledge Systems
Retrieval-augmented pipelines that turn decades of undocumented mainframe code into living, queryable architecture and design documentation.
AI-Assisted Modernization
Legacy reverse engineering with AWS and Claude AI — from COBOL modules to migration plans and Java modernization stories.
LLM Architecture
Model and system design with local and hosted stacks — llama.cpp server, Ollama, Hugging Face, OpenAI — plus the data pipelines that feed them.
SELECTED PROJECTS · OPEN SOURCE
ODEN
A bilingual LLM architecture with hierarchical external memory, adaptive multi-modal attention, mixture-of-experts layers and quantization-aware training.
Universal Data Processor
Ingests CSV, JSON and API sources and turns them into LLM training-ready data through automated schema analysis, cleaning, deduplication and export.
OmegaAI
Hybrid dense + sparse retrieval, persistent memory, distributed training coordination and self-reflective learning for context-aware responses.
quickscrape-dashboard
A configurable, scheduled web-scraping engine with a live Chart.js dashboard for real-time monitoring.
04 · THE SOUND INTERLUDE
15 mantras,one resonance.
An independent Sanskrit mantra album — 15 original tracks blending sound design, spirituality, and AI-assisted creative workflows into meditative, narrative soundscapes.
05 · THE GENOME LIBRARY
Five strands,one helix.
Enterprise & Mainframe
Artificial Intelligence
Cloud, Web & Interactive
Leadership & Delivery
Certifications
06 · BEGIN THE NEXT SEQUENCE