Scale & performance
- High-throughput APIs
- Go profiling & allocation analysis
- Latency and capacity analysis
- Caching strategy
Raj Aryan · Backend & distributed systems
Systems that keep their shape under real-world traffic.
I design and operate high-throughput backend systems where milliseconds, failure modes and infrastructure cost directly affect product outcomes.
Open to senior backend, platform and selected architecture-consulting conversations.
Selected operating scale
Evidence, not decoration.
A few outcomes from production systems where performance and reliability are product concerns.
4.3M+
requests per minute operated across high-scale backend systems.
7M+
Kafka events processed daily in an event-driven pipeline.
500 → 240 ms
location-search latency after a service redesign.
93%
reduction in database-query volume through caching improvements.
≈3%
order-rate improvement from real-time promotion personalisation.
Selected engineering work
Systems thinking under real constraints.
01 / REAL-TIME DECISIONING
Promotion personalisation at runtime
Eligibility, budget awareness and experiment safety brought together in a high-volume decision path.
02 / SEARCH INFRASTRUCTURESearch and discovery across regions
Geo-aware indexing and query paths built for fast responses, safe fallbacks and operational clarity.
03 / RELIABILITYMaking retry amplification visible
A failure-mode analysis of layered retries, dependency pressure and the controls that stop a cascade.
04 / PERFORMANCE & COSTProfiling toward a smaller bill
Allocation, cache and query work that reduced database load while improving system efficiency.
Engineering focus
Start with the problem, then choose the technology.
Reliability
- Failure-mode analysis
- Retries, timeouts & backoff
- Graceful degradation
- Observability & incident analysis
Data & event systems
- Kafka pipelines
- Redis and cache boundaries
- DynamoDB and SQL systems
- Stream processing
Search & discovery
- Solr, Lucene & Elasticsearch
- Geo-sharding
- Indexing pipelines
- Fallback search paths
Infrastructure
- AWS and containers
- Service communication
- Deployment safety
- Cloud-cost optimisation
Product systems
- Promotions and rewards
- Personalisation
- Search, homepage & location
- Business-aware trade-offs
Selected engagements
A focused second pair of eyes on a system that matters.
Backend architecture risk review
A structured walkthrough of the decisions most likely to constrain scale or reliability.
- Scalability-risk assessment
- Reliability gaps
- Prioritised recommendations
- Written action plan
Production reliability audit
Trace how dependency failures become customer impact—and identify controls that break the chain.
- Retry and timeout review
- Failure-mode map
- Observability gaps
- Degraded-mode strategy
Performance & cost review
Find the hotspots worth measuring before a performance or cloud-cost problem becomes structural.
- CPU and memory hotspots
- Query and cache efficiency
- Throughput bottlenecks
- Measurement plan
Writing
Notes from building systems that must keep working.
Retry Storm: How A Single User Crashed 30 ECS Tasks
A production debugging story about infrastructure-level retry amplification and 503 ambiguity.
Read on MediumWhen Apache Solr’s Replication Handler Went Rogue
A production debugging story about replication timing, stale commit points and leader changes.
Read on MediumCache boundaries are an API design decision
How to reason about cache ownership, invalidation and degraded reads.
Read draftLabs, earlier builds & open source
Small experiments with a life beyond the day job.
A small set of projects from Raj’s public GitHub work—useful context alongside the production systems above.
Career Copilot
An AI-powered job-search pipeline for coding agents, built as a public template and workflow.
View repositoryVintage Pacman in Flutter
A mobile take on Pacman that was also featured by Flutter Awesome.
View repositoryHindi Text-to-Speech
A phoneme-database approach to Hindi text-to-speech, shared as a public Python project.
View repositoryGitHub popularity. Snapshot recorded 29 July 2026. Personal GitHub · Work GitHub
About
Calm systems are built by treating constraints as first-class inputs.
“The interesting work starts where a clean happy path meets real traffic, imperfect dependencies and an operating budget.”
Raj is a backend engineer focused on distributed systems that serve product teams and customers under load. The work spans promotions, rewards, subscriptions, search, homepage, dish and location platforms.
A practical next step
Have a backend problem that is too important to guess at?
Share the system, scale and outcome you are working toward. I’ll respond when the problem is a strong fit.