Rehan Akbar
Rehan Akbar

Field Notes & Architecture Log

Rehan Akbar
LocationSingapore
Experience7+ Years
Primary Stack.NET · Node · Cloud
Open for select advisory

Building resilient backends.
Grounding AI in enterprise engineering.

I'm Rehan Akbar, a Senior Enterprise Engineer based in Singapore. For the past seven years, I've designed, modernized, and operated high-concurrency systems across higher education, public sector identity platforms, financial feeds, and global telecommunications.

My core engineering revolves around .NET 8 / ASP.NET Core, Node.js, and cloud-native Azure & AWS architectures. I specialize in the unglamorous mechanics that keep enterprise systems reliable: sub-100ms real-time messaging hubs, execution plan query optimization, concurrent in-memory queues, and Clean Architecture domain boundaries.

As AI reshapes software development, I actively apply these same backend principles to Applied AI & Agentic Workflows — orchestrating multi-agent state machines with LangGraph, designing secure hybrid RAG with Qdrant, and engineering self-hosted n8n automation pipelines with strict observability and PII safeguards.

“This blog is my technical lab notebook — written for engineers who care about real trade-offs, performance boundaries, and making software that holds up in production.”


System Architecture

How I Approach Software Engineering

The Enterprise Core

7+ Years Scaling Production Systems

Deep roots in high-throughput backend architecture, resilient cloud engineering, and database execution tuning.

  • Distributed microservices & Clean Architecture in .NET 8 and Node.js
  • Real-time event hubs with SignalR & Redis Pub/Sub sustaining sub-100ms latency
  • Database performance profiling across SQL Server, PostgreSQL, and Oracle
  • Enterprise security, APIM gateway policies, OAuth 2.0 / Entra ID, and zero-defect CI/CD

The Applied AI Frontier

Bridging Probabilistic Models with Enterprise Rigor

Bringing distributed systems discipline to autonomous agent workflows, state machines, and secure retrieval pipelines.

  • Multi-agent orchestration engines using LangGraph with strongly-typed state transitions
  • Dynamic skill loading and autonomous feedback loops (self-healing test agents)
  • Enterprise RAG architectures combining Qdrant semantic search and BM25 with RRF
  • AI governance: real-time PII masking, hallucination scoring, and Langfuse tracing

Editorial Standard

Guiding Principles of This Blog

What you can expect from every technical deep dive and build log published here.

01

No Toy Demos

Every architecture discussed here considers real-world friction: network failures, concurrency limits, rate budgets, and edge cases.

02

Telemetry Over Intuition

Profiling before patching. Measuring query execution plans, memory allocations, and p99 latency before proposing architectural changes.

03

Deterministic AI Design

Treating LLMs not as magic boxes, but as probabilistic microservices requiring structured schemas, circuit breakers, and evaluation suites.

04

Domain Invariants First

Structuring code around long-lived business logic and Clean Architecture principles that easily outlive front-end framework rewrites.

Toolkit

Technologies & Core Tools

The ecosystem powering my production systems and applied AI experiments.

Languages & Runtimes

  • .NET 8 / C#
  • Node.js
  • TypeScript
  • Python

Cloud & Infrastructure

  • Microsoft Azure
  • Amazon Web Services
  • Docker
  • Redis

Databases & Storage

  • SQL Server
  • PostgreSQL
  • Oracle

Frontend & Interfaces

  • React
  • Next.js
  • Angular
  • Nest.js
Advisory & Consulting

Technical Architecture & Engineering Advisory

Alongside writing, I partner with engineering teams on selective technical engagements: .NET / Node.js backend architecture reviews, cloud concurrency diagnostics, database query profiling, and designing production-grade Applied AI & RAG architectures with strict security and evaluation plans.

Based in Singapore · Available for select advisory engagements

Start a conversationUsually replies within 24 hours

Connect & Follow