┌─────────────────────────────────────────────────────────────────┐
│ │
│ func main() { │
│ me := Engineer{ │
│ Name: "Manjeet Singh", │
│ Location: "Melbourne, VIC 🇦🇺", │
│ XP: "8 years", │
│ Mission: "Build systems that think", │
│ } │
│ me.Build(DistributedSystems, LLMAgents, OpenSource) │
│ } │
│ │
└─────────────────────────────────────────────────────────────────┘
"I didn't just write code. I designed conversations between machines."
Started as a backend engineer building chatbots with IBM Watson. Moved to designing real-time architectures for millions of concurrent players in Ludo King™ & Carrom King™. Then fleet systems, distributed microservices, and serverless AWS at scale.
Today? I'm at the intersection of distributed systems and LLM-powered intelligence — engineering agents that reason, parse, and understand code itself.
The through-line across 8 years: systems that don't just process data, but make sense of it.
Concurrency. Performance. Simplicity.
Worker pools · errgroup · singleflight · channels · atomic ops
When speed-to-ship matters.
I don't just use AI. I build infrastructure for it.
Claude Code · GPT-4 · Llama · Agent Orchestration · AST → Knowledge Graphs
Where the rubber meets the road.
type CurrentFocus struct {
Project string // "Developing Harness"
Stack []string // ["Go","TypeScript", "Neo4j", "AST Parsing", "Claude API"]
Stage string // "Learning Stage"
Goal string // "Give LLMs deep codebase understanding"
Learning []string // ["Claude Certified Architect", "Agentic Patterns"]
}- 🧠 Go Code Analysis Engine — Ingests entire repos, parses Go ASTs, stores relationships in a Neo4j knowledge graph to give LLMs real context about codebases
- 🎓 Claude Certified Architect — Studying agentic architecture, orchestration patterns, multi-agent systems
- ⚙️ AI Agent Infrastructure — Exploring how to build reliable, observable agent systems in Go
|
Dimiour · 2021–2024 Fleet management system built from scratch. Real-time vehicle tracking, matchmaking engine pairing users with optimal vehicles, fuel analytics.
|
GlobalLogic · 2025 Repo ingestion pipeline → Tree-sitter AST parsing → Amazon Neptune knowledge graph → LLM context layer. Enables AI to understand code, not just read it.
|
|
Gametion · 2019–2021 Tournament engine for millions of concurrent players. Bracket generation, real-time state sync, concurrency-safe Redis operations via Lua scripting.
|
smartData · 2017–2019 Cloud-native backend for ingesting wearable + clinical data. High-throughput pipelines with Fitbit, Apple Watch SDK integrations.
|
| Principle | How I Apply It |
|---|---|
| 🔩 Concurrency is a design problem | Think in goroutines, channels, and backpressure from day one |
| 📊 Observability is not optional | AWS X-Ray, structured logs, distributed tracing baked in |
| 🤖 AI amplifies, not replaces | Use Claude Code to think faster, but own every architecture decision |
| 🌊 Events over polling | EventBridge, SNS, CQRS — systems should react, not ask |
| 🗺️ Graphs reveal what tables hide | Knowledge graphs for relationships, not just foreign keys |
"The best systems are the ones that make complexity invisible."