Project OverviewWe are building a high-performance, low-latency trading engine designed for microstructure-based execution strategies in a high-tax (STT) environment.This is NOT a basic retail trading bot.This system requires advanced system-level engineering, multi-core CPU architecture control, shared memory communication, and real-time observability dashboard.The focus of this project is minimizing latency between signal generation and order execution while maintaining regulatory compliance (Order-to-Trade Ratio constraints).The developer must understand low-level performance optimization, concurrency architecture, and Linux system behavior.Core Technical RequirementsPython Version (Mandatory)The engine must use:Python 3.13 Free-Threaded build (3.13t)NOT standard Python 3.10–3.12Reason:Standard Python uses the Global Interpreter Lock (GIL), which blocks true parallelism. In low-latency systems, a 1–2ms delay caused by GIL contention is unacceptable.Multi-Core Architecture with CPU Core PinningThe engine must:Assign specific modules to specific CPU coresUse os.sched_setaffinity (Linux only)Prevent OS core migration (avoid context switching)Modules include:Sentinel (Risk & OTR monitoring)Sonar (Market entropy / regime detection)Oracle (Signal calculation loop)Execution Engine (Order placement)The goal is to eliminate unpredictable latency spikes caused by OS scheduling and cache invalidation.Inter-Process CommunicationStandard Python queues are NOT acceptable.Communication must use:multiprocessing.shared_memoryMemory-mapped buffersLock-free ring buffer architectureReason:Standard queues introduce locking and object allocation overhead, increasing latency.The target is sub-millisecond internal communication between signal generator and execution engine.Latency MeasurementThe system must measure:End-to-end order placement latencyRound-trip time (RTT)Module processing timeUsing:time.perf_counter_ns()Latency histogram loggingThis data must be streamed to the dashboard.Order Execution LogicThe system should:Prefer passive limit ordersInclude 200ms cancel logicManage Order-to-Trade Ratio (OTR)Implement controlled order flooding logic (compliant with broker rules)This is not a simple market order bot.FRONTEND REQUIREMENTS (React Dashboard)The frontend is NOT a trading UI.It is a real-time monitoring and control cockpit.Preferred stack:React (Vite or Next.js)WebSocket for live streamingLightweight charting (Canvas or WebGL-based)Required Dashboard ModulesSentinel PanelReal-time RTT graph20ms lockdown threshold indicatorCPU usage per pinned coreEmergency statusSonar PanelMarket regime indicator (Attack / Veto mode)Entropy score displayZero-trust gate statusOracle PanelWeighted Order Book Imbalance (WOBI) heatmapLiquidity imbalance %Signal strength scoreMust use high-performance rendering (Canvas, not heavy SVG).Execution PanelNet Expected Value (NEV)Fill rate %Cancel rateOrder-to-Trade Ratio (OTR) statusEmergency Kill SwitchDashboard must include:Global kill switchSends signal to monitoring serviceMonitoring service writes flag to shared memoryEngine halts immediatelyDashboard must NOT communicate directly with broker API.Deployment RequirementsLinux-based environment (Ubuntu preferred)Dockerized setup preferredSeparate processes:Trading engineMonitoring microserviceReact frontendGoogle Cloud compatible.10 MOST IMPORTANT SKILLS TO ADDAttach these skills on Freelancer:Python 3 (Advanced Concurrency & Multiprocessing)Must understand GIL, free-threaded builds, shared memory.Low-Latency System DesignExperience reducing microsecond-level bottlenecks.Linux System ProgrammingKnowledge of CPU affinity, process scheduling, performance tuning.Multithreading & Multiprocessing ArchitectureDesigning multi-core optimized applications.Memory Management & Shared Memory IPCExperience with mmap, shared memory buffers.Financial Market Microstructure KnowledgeUnderstanding order books, liquidity imbalance, passive vs aggressive orders.WebSocket & Real-Time StreamingRequired for live dashboard data.React.js (Performance-Optimized UI)Real-time data rendering without UI lag.Performance Profiling & BenchmarkingMust measure and optimize latency.Cloud Deployment (Google Cloud / Linux VM / Docker)Production-ready deployment experience.VERY IMPORTANTAdd this to filter weak developers:Applicants must answer the following:Have you worked with Python shared memory or mmap before?Have you implemented CPU core pinning on Linux?How would you measure internal engine latency?How would you prevent dashboard from affecting trading engine performance?This will eliminate 80% of generic bot developers.