I am a performance-obsessed software engineer with 9+ years of experience building high-scale transactional systems. Currently architecting microservices at Tata Consultancy Services and diving deep into Systems Programming with Go.

I specialize in:

• High-Throughput Systems: Optimizing SQL and concurrency models to handle 1M+ daily events.
• Distributed Architecture: Designing fault-tolerant services using microservices, Kafka, and GCP.
• Low-Level Engineering: Building standard library-only tools to understand OS internals.

Core	Infrastructure	Data & Streaming	Observability

Microservices, Event-Driven Architecture, & Performance Tuning

A cloud-native ecosystem designed to decouple business logic into independent services (Session, Trend, Users).

• The Architecture: Guarded by a custom Reverse Proxy Gateway built with Go's net/http and httputil.
• The Challenge: GORM reflection was causing high latency on write-heavy paths.
• The Fix: Migrated to Raw SQL, reducing P99 latency from ~100ms to 42ms.
• Key Tech: Go, Docker, PostgreSQL, Cloud Run, sync.Mutex Rate Limiter.

AI Infrastructure, Vector Search, & Latency Optimization

• **A cost-control firewall designed to intercept LLM traffic and serve responses from memory using vector similarity.
• **The Architecture: A "Dual-Path" retrieval engine that checks for Exact Matches (O(1)) and Semantic Matches (HNSW Index) before calling OpenAI.
• **The Challenge: Production LLM queries were averaging ~3,000ms latency and incurring high token costs for repetitive questions.
• **The Fix: Implemented Redis Vector Search with Cosine Similarity, reducing P99 latency to <50ms (~5,000x speedup) for cached hits.
• **Key Tech: Go (Goroutines), Redis Stack (RediSearch), OpenAI Embeddings, Docker.

Systems Programming, OS Internals, & Kernel Interfaces

A POSIX-compliant shell built from scratch to master low-level system calls (fork, exec, wait).

• Key Features: Implements non-blocking pipelines (|) and I/O redirection (>, >>) by manually manipulating File Descriptors.
• Why I built it: To understand the abstraction layer between the Go runtime and the Linux Kernel.

Results from productivity-planner load testing (via hey)