Principal Engineer · Systems Architect · Engineering Manager
Raghunath Manyam
I architect AI agent pipelines, full-stack apps and data platforms for businesses that move real money.
Open to Principal Engineer, Systems Architect, Staff Engineer and Engineering Manager roles in regulated industries: insurance, finance and mainframe modernization.
Billionsof dollars in treaty exposureon systems I designed
70+engineers ledUS, India and Europe
<5 ssearchfrom 3–5 minutes
8AI stagesin production
22yearsregulated industries
Three projects to know
AI · 2025 – now
in production
8 Claude stages and one human gate: COBOL in, AWS code out.
An agentic pipeline built as an MCP server. Each stage runs the cheapest Claude model that does the job well, about 35% cheaper than Opus everywhere. The stages read legacy DB2 schemas through my read-only mcp-sql server.
Data · 2020 – 2024
23 tables · 3 jobs
Mainframe to Aurora, plus event-driven reporting.
History moved once with AWS DMS while daily work kept running. Three generalized Glue jobs (extract, transform, load) replaced the ~69 per-table jobs the obvious design needed, feeding reporting for about 50 finance users.
Full-stack · 2020 – 2024
<5 s search
React and serverless TypeScript app where finance clears treaty cash.
Search went from 3–5 minutes to under 5 seconds on 100K+ rows. Results are saved as a DynamoDB snapshot, and every write carries the snapshot's version token, which keeps duplicate clearances at zero.
Technical range
- Claude · MCP · agentic pipelines
- AWS (DMS · Glue · Lambda · Aurora · DynamoDB)
- Mainframe (COBOL · PL/I · IMS-DB · DB2)
- Python · TypeScript · Next.js
- Informatica · PySpark · Teradata
- ITIL · Agile/Scrum (CSM)