I need an end-to-end, AI-driven algorithmic trading system that runs seamlessly in both the US and Indian equity markets. The bot’s sole trading style is scalping, so execution speed, tight risk control, and reliable real-time data handling are non-negotiable. Key scope • Broker connectivity – native API integration with Robinhood for my US account and Kite (Zerodha-India) for my NSE/BSE activity. If you can architect the code in a modular way, future add-ons such as TD Ameritrade or E*TRADE should drop in with minimal refactoring. • Core engine – Python is preferred, but I am open to another performant language if you justify the latency gains. The engine must ingest live market feeds, run the AI model, fire orders, and monitor positions automatically. • AI layer – train and deploy a lightweight model that identifies micro-moves suited to scalping (seconds to a few minutes). Reinforcement learning or an LSTM-based approach is fine as long as it stays computationally efficient and can retrain on fresh data without downtime. • Risk management – dynamic position sizing, per-trade stop-loss, max daily drawdown, and circuit-breaker logic. • Back-testing & paper trading – one click to switch between historical simulation, paper mode, and live deployment. • Dashboard – concise web or desktop UI that shows P/L, open orders, latency stats, and model health in real time. • Documentation – setup guide, annotated source code, sample config files, and short video walkthrough. Acceptance criteria 1. I can connect both broker accounts, start the bot, and see live orders executed within the defined latency budget. 2. A 30-day back-test on Nasdaq and NSE tick data runs without errors and outputs the trade log and equity curve. 3. Switching to paper mode routes orders to the respective paper endpoints and no real trades are placed. 4. All risk limits halt trading automatically when breached and send me an email or Telegram alert. 5. Code passes a peer review style lint/test suite I will provide at hand-off. If you have prior experience building high-frequency or AI-based trading systems, and you can demonstrate low-latency broker integrations, I’d love to see a short demo or repo sample along with your proposal.