Economist, quantitative researcher, and product builder. I spent four years in Model Risk at Deutsche Bank evaluating what models actually establish and where they fail, then moved into venture investing, founded a relationship-research product, and began building research systems end to end.
28 featured models & apps · 62 AI agents built · 3 patented model families
Everything below was built or turned into a working, inspectable public project within one concentrated year. The collection spans econometrics, AI agents, sales systems, finance, medicine, climate research, biocomputing, fusion, quantum error correction, energy, lunar resources, economic history, and aerospace simulation.
Explore the visual portfolio → omrigalor.com
These projects lead with a concrete scientific or engineering question and an inspectable model. The order matches the visual portfolio.
| Project | What to explore |
|---|---|
| Global Human Development Simulator | Test how migration, trade, technology, and household choices shape long-run human development. |
| Closed-Loop Bayesian Battery MRI & Experiment Design Engine | Design the next battery test to distinguish hidden causes of damage. |
| Organoid Intelligence & Silicon Resilience Benchmark | Compare brain-inspired software and silicon-style AI under the same memory disruption. |
| Adaptive Volumetric Light-Field Synthesis | Design one holographic light pattern that reconstructs an object across several depths. |
| Adaptive Lunar Mining & Additive Construction Systems Model | Match Moon-soil processing to local material and test the choice on unfamiliar compositions. |
| Constraint-Aware Fusion Burn & Plasma Stability Control Model | Control plasma position and burn temperature together under disturbances. |
| Exact Topological Quantum Error-Correction Threshold Laboratory | Check whether a correction actually protects quantum information, beyond clearing error warnings. |
| Reusable Rocket Booster Landing Intelligence | Test rocket landing controls against difficult winds, target misses, and fuel limits. |
| Project | What to explore |
|---|---|
| Drag-Aware Vehicular Airflow Recovery Intelligence | Find the narrow conditions where recovered airflow beats its own drag |
| Stratospheric Aerosol Research Risk Model | Put governance and tail risk inside the model |
| Marine Cloud Albedo Research Simulator | Test whether a modeled cloud signal can be detected responsibly |
| Route-Adaptive Mobile Carbon Capture Systems Model | Co-design capture hardware, storage and depot timing on real routes |
| Hybrid Lunar Propellant, Oxygen & Metals Systems Model | Combine polar water ice and molten regolith into one mission-output ledger |
| Longevity Intervention Evidence Engine | Rank longevity hypotheses without hiding translation risk |
| Stochastic Human-Motion Web Extraction Engine | Extract structured web data through modeled human browsing |
| Grounded Compliance Decision-Agent Training & Evaluation Engine | Teach an agent when to answer, clarify or escalate from exact policy clauses |
| Intra-Ancestral Compatibility Engine | Understanding long-term compatibility |
| Employee Attrition Predictor | Explore a hypothesis behind employee turnover |
| Project | What to explore |
|---|---|
| Autonomous Venture Sourcing & Thesis Intelligence Engine | Turn public company evidence into an auditable deal pipeline |
| Systemic Contagion & Optimal Capital-Intervention Engine | Trace cascades and place limited capital where it prevents the most damage |
| AI Investment Diligence Workspace | Turn company materials into investment questions |
| Startup Investment Ranking Simulator | See what drives an investment ranking |
| Credit-Risk Predictor | Investigate a quantitative credit-risk hypothesis |
| Project | What to explore |
|---|---|
| Real-Time Sales Conversation Copilot | Live guidance for what to say next on a sales call |
| Autonomous BDR Research & Outreach Intelligence Brain | Research accounts, map pain to products, and draft the next outreach move |
| Sales Account & Product Propensity Model | Rank accounts and product fit before outreach |
| Pre-Call Account Intelligence & Discovery Planner | Turn sourced account evidence into the next discovery conversation |
| Enterprise Workflow AI Agent Portfolio | Production-style agents for real operating workflows |
Earlier tools and econometric studies remain available on GitHub.
| Repository | Model |
|---|---|
| Cultural Distance Outcome Models | Standalone quadratic specifications for relationship and family-outcome research. |
| Augmented Synthetic Control | Builds a counterfactual from weighted untreated units and tests the result with placebos. |
| Staggered Adoption Lab | Estimates event studies when cities, firms, or countries adopt a policy at different times. |
| Structural Auction Lab | Recovers latent bidder values from first-price bids and tests reserve-price counterfactuals. |
| Grid Stability Control | Tests constrained battery response against generation-loss frequency disturbances across a multi-area grid. |
| Fintech Sales Propensity | Ranks fintech prospects across fit, win likelihood, and value while retaining evidence and uncertainty. |
| Calorie & Deficit Planner | Turns a goal and daily activity into a local-only calorie, macro, and time-to-goal planning view. |
- Provenance over output. A score should trace back to the evidence that produced it.
- Calibration is part of the deliverable. I publish where a model ranks reliably and where it cannot support an individual claim.
- Adversarial validation. Independent checks, placebo tests, and counterfactuals are used to expose fragile conclusions.
- Limits stated up front. Research prototypes, simulations, and illustrative medical or aerospace systems are labeled as such.