Jay846 Jay846

Jay Salvi — Quantitative Researcher | AI enthusiast

Quantitative & AI researcher with specializing in low-level execution platforms, automated feature engineering, and blackbox_os architectures.

SSRN Pre-print | Published: June 23, 2025
📊 Impact: 5,000+ Views | 540+ Downloads | >10% Conversion Rate
Asymmetric Hidden Markov Modeling of Order Flow Imbalances for Microstructure-Aware Market Regime Detection

Zenodo Pre-print | Published: July 9, 2026 (v1)
Process Tool Collapse check for Reliable LLM Agents at Scale (DOI: 10.5281/zenodo.21274028)
📊 Impact: 100+ Views

Zenodo Pre-print | Published: July 17, 2026 (v2)
Process Templates, Sandboxed Code, and Modular Orchestration for Reliable LLM Agents at Scale (DOI: 10.5281/zenodo.21413144)
(Note: v1 published as "Description Engineering Mitigates Tool-Selection Collapse in Large Language Model Agents" - DOI: 10.5281/zenodo.21274028)

Open Source Contributor: NVIDIA CUDA C++ Core Libraries (CCCL) .
Competitions: 1st in Finanza live trading (21% return/3 days) | Top 2.5% in WorldQuant IQC (78k entrants) | DRW Crypto Challenge Top 24%.

Project	Description	Tech Stack
Blackbox-OS	Graph-based agentic operating system for quant workflows featuring LangGraph-orchestrated sub-graphs, self-healing code sandboxes, and adaptive validation guardrails.	Python, LangGraph, pandas, pytest
Neuro-Symbolic-Hypothesis-Engine	An autonomous quantitative discovery agent that extracts governing mathematical laws from raw empirical datasets.	Python, Firecrawl, SymPy, SciPy
Zerodha-Toolkit	Modular toolkit for real-time signal generation, PnL dashboards, and automated execution. Reduced latency by 15%.	Python, pandas, Zerodha API
Regime-Based-Volatility-Model	Backtest and partial implementation of a live-traded GARCH-HMM volatility regime-switching strategy.	Python, Backtrader, Zerodha API
Asymmetric-HMM-OrderFlow	Code companion to SSRN paper: Entropy-weighted Order Flow Imbalance (OFI) + asymmetric HMM for liquidity regime detection.	Python, hmmlearn, scikit-learn
DRW-Crypto-Kaggle-Submission	End-to-end solution for predicting short-term cryptocurrency price changes using tick order book and trade features.	Python, XGBoost, LightGBM
Quant-Research-Project	Volatility pairs trading on Nifty/BankNifty index options using dynamic Kalman Filter hedging. Sharpe 2.19.	Python, PyKalman, Matplotlib

Languages: Python, C++, SQL
Libraries: Pandas, Backtrader, VectorBT, scikit-learn, PyTorch, modelling-imports, SciPy
APIs & Infrastructure: Zerodha Kite Connect, LangGraph, Git, Docker