Case board 01 · systems / writing / life / contact

Last mapped · 2026

00 / Profile01 / Systems + Work02 / Field Notes · WIP03 / Life04 / Contact

Profile / skill preview

Backend systems engineer

  • Java · multithreading
  • Kafka · PostgreSQL
  • Agentic AI · JMH
Asu Singh
Asu / system owner

Senior Software Engineer

Working thesis

I make complex systems behave.

20×trade throughput
90%downtime removed

XTP → measure → improve

throughput · cache · recovery

p99JMH study

Alazium / logs

100 GB+ parsed · triage ↓
tradeXTPJavacacherecoverEODresumemeasurecomparetunep99

Field notes / In pencil

Work in progress.The notebook will open when the writing is ready →

Off the clock / Life

Football & books.Anime · games · curiosity.

Contact / Open channel

Bring me the hard problem.Systems · reliability · AI

Real-time systems · recovery paths · measurable outcomes

Click once → unfold
Double click → dive in
Double click this chit ↗
dive into its board
Tap → unfold
Explore board → dive in

Follow the thread

Case board / 00

Profile / System Owner

Engineer behind the board

A senior software engineer focused on making complex production systems understandable, recoverable, and fast.

Evidence / 01

Core craft

Backend systems

Java, concurrency, distributed workflows, and performance-sensitive services.

Evidence / 02

Operating principle

Trace the evidence

Start with observable behavior, follow the failure path, and measure the recovery.

Evidence / 03

Current direction

Agentic AI

Building useful agents with production-grade constraints, tooling, and guardrails.

Follow the thread across the board →

Case board / 01

Systems / Work

Evidence from production

High-throughput systems, deliberate failure paths, and recovery that can be measured.

Evidence / 01

Low-latency trading

20× throughput

Concurrency, partitioning, and backpressure without losing ordering.

Evidence / 02

Reliability engineering

90% less downtime

Recovery paths, replayable queues, and safer operational defaults.

Evidence / 03

Agentic AI systems

Tools with guardrails

Observable agents designed to operate inside real production constraints.

Follow the thread across the board →

Case board / 02

Field Notes / In progress

The notebook is still in pencil

A space for observations that are being worked into writing, with no publication date promised yet.

Follow the thread across the board →

Case board / 03

Off the Clock / Life

The human side of the map

Books, unfinished ideas, visual observations, and the side quests that make the systems work better.

Evidence / 01

Reading pile

Books + marginalia

Notes from systems thinking, history, design, and good fiction.

Evidence / 02

Side quests

Small experiments

Tools, sketches, and curious prototypes without a roadmap.

Evidence / 03

Elsewhere

Places + perspective

Street-level observations and the ideas that travel back to the work.

Follow the thread across the board →

Case board / 04

Contact / Open Channel

Bring me the hard problem

For backend systems, production reliability, and agentic AI work, open a channel directly.

01 / Direct line

Leave a field note

A few useful details are enough to start.

The board stays open.

Back to the overview ↑Open overview