Expertise
A broad backend toolkit, applied where scale, reliability and change actually matter.
The focus is backend engineering, distributed systems, financial platforms, performance troubleshooting and the practical work required to keep production systems healthy.
Backend engineering
Java and Spring-based backend services, REST APIs, microservices and enterprise application development.
Distributed systems
Event-driven flows, asynchronous work, caching, idempotency and partner integrations where consistency and throughput matter.
Financial platforms
Payments, cashback, lending, NPS and referral programs, plus the configuration and partner onboarding needed around them.
Performance
JVM tuning, thread-dump analysis, memory analysis, query optimization, cache optimization and capacity planning.
Data & storage
Relational persistence and application data access in enterprise environments.
Observability
Production visibility that helps answer “why is this slow?” rather than simply producing more dashboards.
Delivery & leadership
Architecture discussions, technical standards, code reviews, mentoring, production support, secure coding, SDLC, Agile and CI/CD.
Development workflow
AI-assisted development, used with a human in the loop.
AI-assisted development is now part of the engineering loop: useful for getting a first draft out quickly, exploring alternatives, reviewing code and handling repetitive work. Architecture, trade-offs and final decisions remain with the engineer.
Tools in the workflow
CLI-based model workflows
Depending on the task, the workflow may use Claude Sonnet, Haiku or Opus, ChatGPT, and open models such as GLM and Kimi.
Production thinking
When something slows down, start with the system, not the guess.
A useful chain to keep in mind: database pressure can become connection pressure, then thread pressure, then latency, then JVM or GC stress. That is why the whole request path matters.
Metrics, logs, dashboards and traces where available.
Traffic, CPU, memory, GC, disk, queries, pools and downstream latency.
Separate the visible symptom from the subsystem creating it.
Optimize, redesign or scale based on what the evidence supports.