Dhruv Agrawal

• Developed full AI perception stack for agricultural robot — instance segmentation, active viewpoint optimisation, 3D spatial tracking via camera-LiDAR fusion, and plant-level risk prediction
• Designed deterministic firmware migration framework with semantic hardware abstraction and simulation-based validation
• Trained 50+ team members in ML fundamentals, computer vision, and edge deployment workflows
• Represented India at the World Robotics Championship 2024

• Led research task phase covering statistical arbitrage, risk modeling, and portfolio construction
• Conducted and presented FX statistical arbitrage research — IISc podium recognition
• Organised fintech workshop (100+ registrations) bridging quantitative methods and applied finance
• Built multi-signal crash forecasting model and regime-aware portfolio optimisation framework

• Core organising team for Manipal Entrepreneurship Summit (MES) 2026 — 30,000+ attendees, coordinating 150+ startups and 1,300+ competition registrations
• Collaborated with 12 E-Cells nationwide to draft and present Bharat Yuva Innovation Policy Recommendations
• Supported strategic partnership outreach and ecosystem expansion initiatives

FX Statistical Arbitrage Research

Market-neutral cross-currency strategy on a 14-year, 5-second frequency FX dataset. Gross Sharpe 3.81, Net Sharpe 1.99 (1 bp), max drawdown −13.6%, profitable in 12/14 years. Walk-forward OOS evaluation. IISc podium recognition.

Hybrid Crash Forecasting Model

Multi-signal Bitcoin crash prediction fusing EGARCH volatility regimes, LPPL bubble diagnostics, sentiment features, and LSTM sequence modeling. ROC AUC 0.97, 26-day average lead time, 89% precision.

Cross-Asset Portfolio Optimization

Regime-aware allocation framework combining Hierarchical Risk Parity with regime switching. 148% return vs 97.5% benchmark, Sharpe improved 0.41 → 0.68. Walk-forward validation with regime-conditional decomposition.

Post-AGI Digital Economy Simulation

Agent-based framework modeling autonomous economic agents in tokenised economies with reputation currencies and tokenized attention markets. 34% reduction in wealth concentration under targeted policy mechanisms.

AI Perception Pipeline (RoboManipal)

Full perception stack for agricultural robot — instance segmentation, active viewpoint optimisation, camera-LiDAR fusion, 3D spatial tracking, and plant-level risk prediction. World Robotics Championship 2024.

Firmware Migration Framework

Deterministic cross-architecture porting using semantic hardware abstraction layers, automated code translation, and simulation-based validation across MCU families.