AiFund‑Lite: RL Hedge‑Bot
Clean PPO/LSTM baseline, modular risk manager, simple backtester, Kraken integration via CCXT (paper mode).
Open RepoI design, train, and deploy trading engines, sentiment snipers, and DeFi experiments. Less theory, more PnL. Berlin-ready.
RL-driven strategy on 1m crypto data. Paper traded with risk controls.
DistilBERT signals for meme pumps. Fake-pump detector with thresholds.
DEX sniping skeleton with orderflow sampling + backoff logic.
*Metrics shown are from controlled backtests or sandbox runs. Repos include notebooks to reproduce.
Clean PPO/LSTM baseline, modular risk manager, simple backtester, Kraken integration via CCXT (paper mode).
Open RepoHeadline sentiment → event-driven signals. Telegram alerts, throttled execution, and PnL logging.
Open RepoLatency first. Pool watcher + MEV-aware backoff. Designed for Raydium/MEXC/Bitget research.
Open RepoDecentralized volatility stabilizer concept. Tokenomics, PPO controller sketch, Grafana dashboards.
Open Repo1m bars, rolling σ filter, position sizing by z-score.
GARCH sketch vs LSTM volatility head – tradeoffs in microstructure.
Joint signal: text polarity + AMM slippage → better pump filters.
I'm Garvit — a self-taught quant/engineer who builds lean, shipping‑first systems. I like brutal simplicity, readable code, and experiments that expose truth. If you want someone who prototypes in days (not quarters) and iterates with real data, let’s talk.
Berlin founders & quant desks: I’m available for internship/working student/contract roles.
CV available on request. References & code walkthroughs on call.