Milan Prajapati
"Building distributed systems, real-time platforms, and scalable backend infrastructure."
Design-focused backend architect with production expertise in event-driven microservices and low-latency infrastructure. Specializing in high-frequency message processing pipelines, GC-optimization patterns in .NET, and scalable telemetry middleware.
ABOUT // INFRASTRUCTURE ARCHITECT
ENGINEERING LEADERSHIP & DISTRIBUTED TOPOGRAPHY
> Architectural Philosophy
My approach to software engineering centers on system reliability and throughput. I believe backend performance is a product feature, not an afterthought. Rather than scaling hardware prematurely, I focus on maximizing the runtime efficiency of existing nodes through cache architecture, zero-allocation data structures, and decoupled message routing.
> CORE DOMAINS
EXPERIENCE // ENGINEERING IMPACT
PRODUCTION ARCHITECTURE & SYSTEMS PERFORMANCE
Led design and implementation of a flagship Realtime RFID Telemetry Platform, managing high-throughput data streams connecting multiple physical warehouse gates with cloud databases.
Focused on system bottlenecks: analyzed heap memory maps, configured thread-pool thresholds, isolated domain boundaries, and designed asynchronous network queues to ensure zero packet drop rates.
Database & API Latency Reduction
99.7% Latency ReductionEliminated synchronous entity tracking loops and N+1 query patterns by introducing targeted Redis indices and disabling Entity Framework change tracking on read-heavy routes.
EXPERTISE // INFRASTRUCTURE METRICS
ENGINEERING CAPABILITIES & ARCHITECTURAL STACK
Distributed Systems
- State Machine Replication
- Partition Tolerance
- Data Partitioning
- Consensus Protocols
Event Driven Architecture
- CQRS Handlers
- Event Sourcing Patterns
- Durable Outbox Pattern
- Idempotent Consumers
Real Time Communication
- WebSockets Transport
- Bi-directional Streaming
- Connection Pooling
- Zero-Allocation Hubs
Performance Optimization
- Gen 2 GC Mitigation
- ArrayPool Renting
- Stack Allocations
- Low-level Memory Profiling
Microservices
- Bounded Contexts
- API Gateway Routing
- Service Registry
- Inter-service RPC / QUIC
Cloud Native Systems
- Docker Compose Orchestration
- CI/CD Actions Automation
- Stateful Sets Configuration
- Load Balancing Topography
Messaging Systems
- Prefetch Tuning
- Dead Letter Exchanges
- Topic Exchanges Mapping
- Backpressure Regulation
Database Optimization
- Index Analysis
- Execution Plan Tuning
- Caching Strategies
- EF Core Query Interceptors
PHILOSOPHY // CORE PRINCIPLES
SYSTEM ARCHITECTURE PHILOSOPHY
Scalability First
Build every system with horizontal scaling constraints in mind. Restrict local memory dependencies, implement decoupled storage, and favor stateless architectures that expand gracefully.
Measure Before Optimize
Never refactor code based on intuition. Use memory profilers, heap dump analyzers, and CPU telemetry to locate precise bottlenecks before writing low-level optimizations.
Event Driven by Default
Isolate services using message queues. Rely on asynchronous event-driven triggers rather than synchronous REST calls to prevent cascading service failure patterns.
Reliability Over Complexity
Reject unnecessary software layers. Choose simple, verifiable abstractions that are easy to monitor and trace rather than complex designs that complicate fault diagnosis.
Performance is a Feature
Enforce strict performance contracts. Treat latency regressions as critical system bugs and design memory pipelines to prevent high-frequency Garbage Collector pauses.
PRODUCTION PROJECTS // ENTERPRISE DEPLOYMENTS
COMMERCIAL CORE PLATFORMS & AGENTS
Bepco RFID
IoT Telemetry & Middleware Platform
Real-time IoT telemetry platform connecting physical RFID hardware readers to a high-frequency .NET middleware layer and a centralized web application. Architected to support multi-site concurrent warehouse logistics.
Parallelized execution in .NET to handle concurrent gate reads, resolving AJAX serialization blocks and reducing read-to-dispatch latency.
Optimized data paths to use Span<T> and stack allocations on high-frequency SignalR pipelines, eliminating garbage collection pauses.
Integrated dual-source event verification using Hangfire-scheduled background processors and AI camera verification.
Layered Redis caching over MySQL databases to maintain low-latency reads as scale expanded to multiple warehouse nodes.
"Personal projects and experimental systems are hosted separately in the CTO Lab."
Access my active playground of P2P network engines, CSnakes-interop trading analytics, and containerized home-lab deployments.