Pramod Thaz

Multi-Asset Analytics · Bayesian Decision Systems · Custom AI for Finance

Austin, TX · CAIA Charterholder · CFA Level III Candidate

Technical Appendix ⬇

Daily Regime Pulse ↗ · Blog ↗

Private repositories · request collaborator access

EDUCATION

MBA, Finance — UCLA Anderson

B.Tech (Hons) — Indian Institute of Technology

CAIA Charterholder · CFA Level III Candidate · GE Leadership Program

PROFESSIONAL SUMMARY

Senior investment professional with 25+ years architecting and directing multi-asset analytics, risk, and decision systems, supporting technology infrastructure across public markets and commercial real estate.

Specialized in three high-value capabilities:

› Custom AI fine-tuning and agentic systems purpose-built for finance applications in CRE and public markets
› Bayesian workflows and probabilistic modeling for portfolio construction, regime detection, and uncertainty-aware decision making
› Institutional multi-asset analytics that integrate liquid and illiquid holdings into coherent performance, risk, and liquidity frameworks

Combines deep domain ownership with modern AI and probabilistic methods. Currently engaging selectively with institutions and platforms seeking these capabilities on an independent or advisory basis.

CORE CAPABILITIES

› Custom AI fine-tuning & agentic systems for finance (CRE & public markets)
› Bayesian portfolio construction & regime detection
› Multi-asset performance, attribution & risk systems
› Hybrid liquid/illiquid data architecture
› Model risk & analytics governance
› Investment-committee narrative & decision support

TECHNICAL FOUNDATION

AI & Bayesian

Custom AI fine-tuning for finance
Claude + OpenCode agentic workflows
PyMC / Bayesian methods

Platforms & Data

Multi-asset performance & attribution
Hybrid liquid/illiquid data
SQL + graph + document stores
Power BI / Power Platform
Bloomberg / FactSet

Languages

Python
C#
JavaScript/ES6
R
Rust (beginner)

ENGAGEMENT FOCUS

Premium independent and fractional work in multi-asset analytics, Bayesian decision systems, and custom AI for finance — targeting allocators, managers and family offices that require institutional rigor with modern probabilistic and AI capability.

Austin, TX
github.com/pramod-thaz ↗
Full professional profile and résumé available on request

CONSULTING ENGAGEMENTS

› IN PROGRESS: A multi-asset portfolio-management workstation built on a FastAPI + DuckDB engine where Python and live-evaluated Excel run as peer compute modules, with live pricing, watchlists, charts, and dashboards driven from markdown. Deal documents are indexed via PDF embeds and transclusion, keeping every screen and calculation directly traceable to, and searchable against, source documents. Designed around a no-durable-plaintext model — analysis runs in-memory and is wiped on exit — suited to compliance-sensitive mandates.
› IN PROGRESS: For any excel based workflow, designing and implementing a nimble, accurate, and fully auditable infrastructure that keeps Excel as the primary analyst interface while leveraging modern data engines (DuckDB, NoSQL/document stores), vector databases, and LLM capabilities. The system must support high traceability, strict compliance, and scalable coverage while being able to operate offline just like native Excel.
› Directed independent CRE underwriting, analytics, and asset-management practice. Led acquisition support for ~$100–300M of multifamily and creative office assets and advised large managers on operational strategy and decision systems.
› Designed and delivered production analytics platforms and data infrastructure for family offices and institutional clients, including full-cycle asset management systems and the first comprehensive budgeting framework for a large SFR platform (~30,000 units / ~$5B).
› Implemented performance-attribution and manager-appraisal frameworks for public pension separate accounts and direct investments, supporting through-the-cycle allocation decisions.

PROFESSIONAL EXPERIENCE

Grafolio.ai — Founder · Apr 2026–Present · Austin / San Antonio, TX

› Grafolio designs, deploys and governs autonomous AI agents for asset managers, pensions and endowments, REITs, and real estate and private credit managers. We start from the work your team already does: the reporting cycle, the investment committee memo, the underwriting model, the compliance review. We turn it into agent workflows that are cited, logged and ready for audit.
› Meridian by Grafolio is the AI-first CRE portfolio platform and reference implementation: Portfolio Graph traversal with line-item citations, driving approved Excel models in place. Details at grafolio.ai ↗.

Teachers Retirement System of Texas — Austin, TX

› Directed institutional performance measurement infrastructure for the ~$200B multi-asset trust, encompassing funds, aggregates, and composites. Established production systems and multi-decade validated historical repository serving as the analytical foundation for research and governance.
› Advanced multi-asset and alternative risk premia research. Researched AI-accelerated multifactor equity portfolio analytics and Bayesian regime-aware analysis using Claude, OpenCode agentic workflows, and probabilistic modeling.
› Translation of complex performance and risk outputs into clear narratives for internal stakeholders.

USAA Real Estate — San Antonio, TX

› Led the enterprise performance measurement and attribution function across all funds, investments, and assets. Institutionalized data accuracy and traceability standards that corrected reported performance by ≥450 bps in key asset classes.
› Drove strategic process redesign that improved both internal decision quality and external reporting credibility.
› Represented the firm in industry standard-setting bodies (NCREIF/MSCI) and chaired performance methodology work.

Irvine Company Apartment Communities — Irvine, CA

› Held primary P&L and operational responsibility for a large multifamily portfolio while leading operational analytics across a ~50,000-unit platform.
› Designed and institutionalized advanced analytical frameworks (goal-based revenue optimization and continuous-time simulation of asset dynamics) that contributed measurable value creation.

GE Capital Real Estate — Norwalk, CT

› Architected the comprehensive risk and analytics framework used to stress-test an $80B CRE portfolio; presented and defended methodology and results to Federal Reserve oversight.
› Directed quantitative underwriting, valuation, and syndication systems supporting approximately $3B of complex equity, debt, and NPL transactions with major institutional counterparties.
› Established model suite for commercial asset valuation, debt pricing, and leveraged structures used across the global portfolio; trained international teams on methodology and application.

Earlier Experience

› Trimont / Lehman — sole asset management of distressed multi-asset portfolio; developed valuation and recovery frameworks.
› IBM / Invensys — software architecture and development management for mission-critical industrial systems; patent holder for extensible process-control architecture.

Pramod Thaz — Technical Appendix

Custom AI for Finance · Bayesian Decision Systems · Multi-Asset Analytics

Austin, TX · CAIA Charterholder · CFA Level III Candidate

CUSTOM AI SYSTEMS FOR FINANCE

› Domain-specific fine-tuning and adaptation of AI models for commercial real estate and public-markets applications.
› Agentic development workflows (Claude + OpenCode) used to accelerate design, implementation, and validation of analytics systems and multifactor equity portfolio models.
› Emphasis on retaining full domain ownership: AI accelerates execution; architecture, assumptions, and decision logic remain expert-controlled.
› Application areas: CRE underwriting and cash-flow systems, multifactor equity models, hybrid public/private analytics pipelines, research prototyping.

BAYESIAN WORKFLOWS & PROBABILISTIC MODELING

› Portfolio construction under uncertainty.
› Regime detection and time-varying parameter frameworks.
› Hierarchical and partial-pooling models.
› Model risk, posterior predictive checking, and decision-oriented uncertainty quantification.
› Gaussian processes and surrogate modeling for complex systems.
› Primary stack: PyMC and related probabilistic programming tools.
› Design principle: explicit uncertainty that improves capital-allocation decisions rather than purely statistical elegance.

MULTI-ASSET ANALYTICS PLATFORMS

› Institutional performance measurement and attribution engines spanning public markets and private/CRE holdings.
› Hybrid data architectures that reconcile high-frequency liquid market data with slower, unstructured private-market and real-estate information.
› Risk frameworks, stress testing, and early-warning systems.
› Liquidity and total-portfolio views across asset classes.
› Production systems with auditability, data provenance, and multi-decade historical integrity.
› Investment-committee translation of complex quantitative outputs into clear decision narratives.

DATA ARCHITECTURE & PLATFORMS

› SQL — production relational systems and performance-critical analytical queries.
› Graph databases and knowledge graphs — entity-relationship modeling for multi-asset and document-linked data.
› Document / NoSQL stores for unstructured private-market and CRE content.
› Power BI and Power Platform — stakeholder dashboards and operational workflow automation.
› Excel–Python production integration for large-scale analytical tasks and institutional twin models.

LANGUAGES & CORE TOOLING

Layer	Tools / Languages	Role
Primary execution	Python	Production systems, modeling, automation
AI / Agentic	Claude, OpenCode, custom fine-tuning workflows	Domain-adapted AI systems for finance
Probabilistic	PyMC and related stacks	Bayesian portfolio & risk frameworks
Data	SQL, graph databases, document stores	Hybrid liquid/illiquid architectures
BI & Automation	Power BI, Power Platform, Excel–Python integration	Decision support and operational scale
Supporting languages	C#, JavaScript/ES6, R	Working proficiency
Emerging	Rust	Beginner
Market data	Bloomberg, FactSet, NCREIF/MSCI standards	Ongoing institutional use

DESIGN PHILOSOPHY FOR PREMIUM ENGAGEMENTS

Systems and frameworks are built to be:

› 1 Decision-grade — outputs that investment committees and CIOs can act on.
› 2 Uncertainty-aware — Bayesian where it materially improves capital allocation.
› 3 Domain-adapted — AI fine-tuned or structured specifically for finance (CRE + public markets).
› 4 Auditable and production-ready — institutional standards of provenance and governance.
› 5 Hybrid-capable — able to span liquid and illiquid assets without forcing artificial uniformity.

Full professional profile and résumé available on request.