About · the person behind the notes

Jagdish Salgotra

Software engineer · systems architect · New Delhi, IN · 15+ years in production.

loc · new delhi, in·exp · 15+ years·writing here since · 2023

I build software where performance, reliability, security, and cost are first-class design constraints. The past fifteen years have been mostly distributed systems - systems that must behave predictably under load, failure, and real-world mess. Concurrency, data pipelines, large-scale production architectures.

Core expertise

  • 01

    Distributed systems & concurrency

    Designing predictable systems under load, latency, and failure - using the JVM and modern concurrency models.

  • 02

    Cloud-native infrastructure

    Building and operating services on AWS and GCP with Kubernetes, containerisation, and cost-aware scaling.

  • 03

    Data & search systems

    Architecting low-latency data pipelines and semantic search with relational stores and OpenSearch.

  • 04

    System design & architectural trade-offs

    Deliberate decisions balancing latency, cost, correctness, and operational complexity under real production constraints.

  • 05

    Production observability

    Metrics, tracing, and alerting designed to surface failure modes before users feel them.

  • 06

    Applied AI with explicit boundaries

    AI systems with confidence checks and operational control - not demos, not vibes.

My story

Over the past fifteen years my work has centered on designing and operating distributed systems where performance, resilience, and operational clarity matter more than novelty. Most of what I write here comes from solving real production problems - slow services, noisy pipelines, scaling bottlenecks, systems that fail in subtle ways.

This site is a personal technical notebook. Deep dives into concurrency, performance tuning, cloud optimisation, distributed systems, vector search, and applied AI. The emphasis is on reasoning about trade-offs, failure modes, and system behaviour drawn from real production work - not tutorials, not trends.

Alongside the writing I built NoteSensei - an AI companion that reads every article and answers questions using only the ideas and language from the blog itself. Intentionally constrained, non-hallucinatory, and designed to challenge reasoning rather than give generic answers.

The goal is simple: share practical, production-grade thinking that helps engineers reason better about complex systems. Whether you’re debugging a slow service, preparing for a senior-level interview, or exploring new concurrency and AI models.

- Enjoy the notes.

Representative stack

what I reach for

Languages

JavaPythonGoC++TypeScript

Cloud & infra

AWSGCPKubernetesTerraformDocker

Data & search

PostgreSQLOpenSearchKafkaRedisClickHouse

Frontend

ReactNext.jsTailwind-adjacent CSS

Applied AI / ML

RAGVector DBsLLM evalsEmbeddings

What I believe in

  • Code is communication

    The best code isn’t just functional - it’s a conversation with future maintainers, most of whom are future you.

  • Understanding over memorisation

    Anyone can follow a tutorial. Real engineering is understanding trade-offs and making informed decisions.

  • Shipping matters

    The perfect solution that never ships is worse than the pragmatic one that works in production.

Let’s connect

Always up for a good engineering conversation.

Opportunities, collaborations, interesting problems - or just to compare notes on something production recently taught you.