Senior Full-Stack Engineer · AI Systems · Algorithmic Trading
I build production-grade trading infrastructure, real-time AI dashboards, and multi-tenant SaaS platforms. I work at the intersection of low-latency systems, LLM orchestration, and full-stack product engineering.
| Domain | What it means |
|---|---|
| Algorithmic Trading | State-machine-driven execution engines with zero-disk-write tick loops, MT5 integration, and AI-powered order flow analysis (CVD, absorption, volume profile). |
| Real-Time AI Dashboards | Streamlit + Next.js command centers unifying live market data, LLM routing, voice assistants, and multi-service telemetry into one cockpit. |
| LLM Orchestration | Multi-provider proxy routers with auto-fallback chains, agent swarms, and Telegram-driven AI gateways. |
| SaaS Platforms | Multi-tenant BSAAS architectures with FastAPI backends, Next.js frontends, PostgreSQL, and Telegram bot automation. |
| Video Intelligence | Vision-based signal generation from chart screenshots and YouTube content credibility profiling. |
Post-News Breakout Trading Engine — Production Grade
A 5-stage state machine (PRE_NEWS_GUARD → CONSOLIDATION_SCAN → BREAKOUT_DETECTION → ENTRY_EXECUTION → POSITION_MANAGEMENT) built for high-frequency post-news breakout trading on BTCUSD.m and EURUSD.m.
- Zero-disk-write tick loop — all state retained in-memory with deque ring buffers
- FabioValentiniScalper — CVD, absorption, volume profile order flow analysis
- FastAPI + WebSocket telemetry — 1Hz real-time state streaming to Next.js dashboard
- 180/180 tests passing — pytest, ruff, mypy strict mode
- Exponential backoff MT5 reconnection — max 10 retries, jitter, degraded mode
- Dual-write journal — local JSONL primary + Supabase mirror
Unified Trading Orchestration Spine
The canonical orchestration layer for a multi-bot trading ecosystem. Owns signal schema v1, strategy mode routing, soft risk gating, and dual-write journaling.
- Signal schema v1 + normalizer — typed, versioned signal contracts across 4+ bots
- Strategy mode router — reads strategy_config.json, switches between regime_btc / ema / scalp modes
- Soft risk gate — pause/resume, symbol allowlist, confidence threshold, concurrency limits
- Hermes resolve — WSL integration for cross-platform agent communication
- CLI interface — 15+ commands for self-test, preflight, journal probing, hive queuing
Unified Trading + AI Dashboard
A 10-page Streamlit application unifying DeepChartsPro, ST4RBOY Core, Hermes, and OmniRoute into a single control cockpit.
- 10 page renderers — home, core_status, terminal, signals, signal_scanner, trading_terminal, system_health, jarvis, settings, logs, omniroute
- ENI voice assistant — browser-native push-to-talk, context-aware responses, dedup logic
- Live market ticker — sidebar price strip with real-time updates
- Command palette — Cmd+K fuzzy search overlay for navigation
- Jarvis theme system — 6 theme presets, particles, focus mode
- WebSocket integration — live MT5 feed, AI brain outputs, signal pipeline
BSAAS — Telegram Automation for South African Businesses
A multi-tenant SaaS platform automating customer communication, bookings, and support through Telegram.
- Full-stack architecture — Next.js 14 + TypeScript frontend, FastAPI + PostgreSQL backend
- Multi-tenant data isolation — SQLAlchemy 2.0 + Alembic migrations
- JWT authentication — secure server-side auth + authorization
- Kibo chatbot — Telegram-native AI assistant with intent classification via OmniRoute
- Production-ready infrastructure — Docker Compose, health checks, seed scripts
24/7 Session-Driven Execution Engine
A production trading bot with session-aware strategy rotation, risk management, and multi-broker support.
- Session router — Tokyo, London, New York session detection with SAST timezone handling
- Strategy rotation — auto-rotates between Tokyo Sweep, London Open Fade, London Close Reversal
- Risk management — margin guard, portfolio heat, correlation matrix, partial close system
- ML validator — machine learning signal validation layer
- News filter — filters trades around high-impact news events
- Heartbeat server — health monitoring endpoint for uptime tracking
- State persistence — crash recovery with position metadata restoration
YouTube Credibility Profiler — Terminal CLI
Analyzes YouTube content creators for actionable substance vs. empty buzzwords using transcript extraction and credential scoring.
- yt-dlp integration — extracts auto-captions + metadata from any YouTube video URL
- Antigravity SDK — optional LLM-powered credential evaluation
- AWS S3 storage — persists transcripts and analysis results
- Poetry-packaged CLI — deep_profiler entry point with pytest + ruff + mypy
Python · TypeScript · JavaScript · SQL · Bash · PowerShell
FastAPI · Uvicorn · SQLAlchemy · Alembic · Supabase · PostgreSQL · pydantic
Next.js 14 · React 18 · TypeScript · Streamlit · Plotly · Tailwind CSS
MetaTrader5 · NumPy · pandas · WebSockets · yt-dlp · boto3
OmniRoute (multi-provider LLM proxy) · Claude · Gemini · DeepSeek · Telegram Bot API
Docker · Docker Compose · Turborepo · pnpm · GitHub Actions · AWS CDK · WSL
pytest · ruff · mypy · prettier · Jest · pytest-asyncio · pytest-cov
- Phase 6-7 autonomous self-healing loops — thread-safe telemetry bridges, LLM fallback chains
- Vision-based signal generation — screenshot analysis → trade signals
- Multi-agent orchestration — Hermes agent swarm with task distribution
- Real-time risk analytics — portfolio heat, correlation matrices, margin guards
I'm always interested in collaborating on production-grade trading systems, AI infrastructure, or full-stack SaaS products.
- GitHub: @JayV42069
- Email: Available on request
Built with obsessive attention to production readiness, test coverage, and clean architecture. No scaffolding, no tutorials — everything here runs live.