╔═══════════════════════════════════════════════════════════════╗
║ ║
║ █████╗ ███╗ ███╗ █████╗ ███╗ ██╗ ║
║ ██╔══██╗████╗ ████║██╔══██╗████╗ ██║ ║
║ ███████║██╔████╔██║███████║██╔██╗ ██║ ║
║ ██╔══██║██║╚██╔╝██║██╔══██║██║╚██╗██║ ║
║ ██║ ██║██║ ╚═╝ ██║██║ ██║██║ ╚████║ ║
║ ╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═══╝ ║
║ ║
║ backend engineer · distributed systems ║
║ real-time data · fintech infrastructure ║
╚═══════════════════════════════════════════════════════════════╝
$ whoami
> aman dugar
$ cat /etc/profile.d/stack.conf
ROLE="Backend Engineer"
FOCUS="Distributed Systems · Real-time Data · Infra"
LOCATION="Bengaluru, India"
TRANSACTIONS_PER_MONTH="100M+"
UPTIME="99.99%"
$ ./describe --current
Building fault-tolerant systems that process money at scale.
Obsessed with correctness, latency, and the code paths nobody sees.const aman = {
role: "Backend Engineer @ Flagright",
domain: ["AML/Compliance", "Fintech", "Risk Infra"],
languages: ["TypeScript", "JavaScript"],
backend: ["Node.js", "Serverless", "Fargate"],
databases: {
analytics: "ClickHouse", // billions of records, single-digit ms queries
document: "MongoDB",
kv: "DynamoDB",
search: "OpenSearch",
},
streaming: ["Kinesis", "SQS", "Lambda"],
infra: ["AWS (9 regions)", "CDK", "Terraform"],
ai_ml: ["Claude API", "RAG pipelines", "Voyage embeddings"],
currentlyBuilding: [
"AML investigation automation engine (aif-evo)",
"AI-powered knowledge bot over private codebases",
"Multi-region transaction monitoring at scale",
],
principles: [
"Measure before optimizing",
"Simple > clever",
"Design for failure modes",
"Correctness is non-negotiable when it's money",
],
} as const;| System | Scale | Stack |
|---|---|---|
| Transaction Monitoring Rules Engine | 100M+ txns/month | Node.js · ClickHouse · DynamoDB |
| AML Investigation Automation | Multi-tenant SaaS | TypeScript · MongoDB · Claude API |
| Multi-Region AWS Infra | 9 AWS regions | CDK · Fargate · Kinesis |
| Sanctions Screening Pipeline | Real-time batch | Lambda · S3 · OpenSearch |
| RAG Knowledge Bot | Private codebase Q&A | Voyage · OpenSearch · Slack |
[ClickHouse] Migrated analytics from MongoDB → ClickHouse, 40× query speedup
[Auth] Rebuilt PropelAuth RBAC V2 — multi-tenant permission system
[Infra] Provisioned 9-region AWS deployment via CDKTF
[AI/ML] Built hybrid search RAG pipeline (Voyage + OpenSearch + Claude)
[Rules] Rewrote V8 rules engine — pluggable, tenant-scoped, streaming-aware
[SAR] Designed SAR/iSTR regulatory reporting pipeline from scratch
[Sanctions] Two-level hit cache, pMap concurrency, checkpoint-based resumption
[AIF-EVO] NarrativeContextBuilder — LLM handles prose, all figures pre-injected
{
"email": "amandugar@gmail.com",
"linkedin": "https://linkedin.com/in/amandugar",
"open_to": ["Staff Eng roles", "Backend Infra", "Fintech/AI infra"]
}┌─────────────────────────────────────────┐
│ │
│ good systems scale. │
│ great systems stay simple while │
│ handling the chaos underneath. │
│ │
└─────────────────────────────────────────┘