Projects

Detailed demos, tech notes, and tools.

All Trading AI/ML Tools Business

MarketLabs — Terminal, Analytics & Test Bench

Role: Lead Engineer Stack: Java • C++ • PostgreSQL Status: In active development
Backtesting Live Feeds Risk Engine Plugins

Unified terminal for real-time quotes, portfolio insights, and a sandbox for rapid idea → test → deploy. Bridges research and execution with low-latency hooks and risk routines.

Live Demo Source Docs
What it does
  • Strategy bench with walk-forward testing & parameter sweeps.
  • Portfolio diagnostics (VaR, drawdown, factor tilt).
  • Execution adapters for paper/live trading.
Why it matters
  • Shortens the loop from research to live deployment.
  • Reduces model risk via built-in sanity checks.
  • Composable: drop in new data sources/models with plugins.
Architecture
  • Modular services for data, backtests, and risk via in-proc message bus.
  • Strategy API for C++/Java models with JNI bridge.
  • PostgreSQL for state; columnar cache for hot paths.
Roadmap
  • GPU-accelerated sims
  • Paper/live trade blotter
  • On-device feature store
Terminal Risk Backtests

Advanced Portfolio Analytics Engine

Stack: Python • NumPy • FastAPI Focus: Risk, return, scenarios Status: Beta
VaR Stress Tests HMM Regimes

Predictive analytics for risk/return, factor attribution, stress tests, forward simulations, and scenario tooling for decision support.

Interactive Demo Repo API Docs
Feature Map
  • Monte Carlo & block bootstrap paths
  • Factor decomposition & tilt guardrails
  • Regime detection with HMMs
Sample Metrics
  • Max DD ↓ 14% on backtests
  • Turnover ↓ 23% with constraints
  • Sharpe ↑ 0.21 vs baseline
Integration

REST endpoints for ingestion, simulation, and reports. Designed to plug into MarketLabs or notebooks.

EquityLens AI

EquityLens AI — Automated Equity Research

Stack: Python • PyTorch • NLP Status: Prototype → Private
NLP Valuation Thesis Gen

Neural + NLP pipeline that ingests SEC filings and market data to surface mispricings, generate valuation models, and produce thesis briefs with risk flags.

Sample Report Model Cards
Modules
  • 10-K parser → entity resolution, KPI extraction, footnote linker
  • Factor regressions & sanity checks for spurious alpha
  • Report composer (export: PDF / Notion / Markdown)
Artifacts
  • Coverage: S&P 500 + ADRs
  • Latency: ~4 min per filing
  • Explainability: SHAP summaries on drivers
PlanWise AI

PlanWise AI — Personal Finance Engine

Stack: Python • StreamlitFocus: Planning
BudgetingTax-AwareForecasts

Retirement planning, budgeting, tax-aware projections, and adaptive plans. Built for clarity and speed.

Open App Whitepaper

Embed your Streamlit app iframe here once deployed.