Principle & Narrative

I build decision systems where production rigor meets agentic leverage.

My work turns ambiguous business questions into pipelines, models, and agentic tooling designed to operate in high-stakes, regulated environments.

Why I work this way

Most of my career has been spent at the seam between analytics and operations — the place where a model either changes a decision or gets ignored. That seam is where I learned to build for production first and optimize for cleverness second.

“The point of a decision system isn’t to be right about the future — it’s to make the organization legible to itself when the future shows up.”

At SoFi that means Monte Carlo anomaly detection across pricing deployments that touch OCC, Internal Audit, and 2LOD. At Autodesk it meant stitching Salesforce, product telemetry, and finance into a single GTM surface that Sales and RevOps could actually self-serve. At ZipRecruiter it was reallocating a $300K/month budget weekly against signal, not opinion.

The new layer is agentic. Claude-powered tools that plan SQL, self-correct, and render visualizations. RAG + MCP scaffolding that lets non-analysts interrogate pricing grids safely. The projects in the Gallery are working demos of that thesis — the case study below is what it looks like when the same thesis lands inside a regulated, high-stakes system.

Senior Impact

TimelineInstitutionRoleCore Impact
May 2022 — Present
SoFi
San Francisco, CA
Senior Data Scientist

Compliance analytics + AI tooling across a $20B+ lending portfolio. Built a RAG + MCP tool layer and Monte Carlo anomaly engine that cut analysis cycle time 70%+ while sustaining 99.9%+ accuracy across 1,000+ pricing deployments.

Read Case Study
May 2020 — May 2022
Autodesk
San Francisco, CA
Senior Analytics, BI & GTM Strategy

GTM reporting infrastructure for Sales, Marketing, and Finance. Automated CAC, funnel velocity, and attribution; shipped predictive ARR + retention models and self-serve dashboards that cut time-to-insight by 70%.

May 2019 — May 2020
ZipRecruiter
Santa Monica, CA
Marketplace Strategy Data Scientist

Owned a $300K/month global marketing budget. Built LTV + retention models and reallocated spend against real-time campaign signal — delivered +5% MAU in Q4 and 25% QoQ growth.

Jul 2017 — May 2019
PayPal
San Jose, CA
Data Engineer

Airflow + SQL pipelines powering RMA and refurbishment analytics. Tableau reporting for business-trend evaluation; ETL automation that cut pipeline development time by 50%.

How I Operate

Production-Grade Rigor

Decisions that touch regulators and $20B portfolios don't get prototypes. I ship Monte Carlo alerting, dbt-governed pipelines, and reviewable experiment diffs — because the floor has to hold before the ceiling matters.

Cross-Functional Fluency

The useful work happens at the boundary. I translate between policy, engineering, and executive narrative — weekly exec reads that set pricing strategy, SOPs that scale execution, and frameworks others can run without me in the room.

Agentic Systems Thinking

LLMs change the shape of what a single analyst can do. RAG over pricing artifacts, MCP tool layers over experiment metadata, self-correcting agents that plan and execute — I design the scaffolding so the leverage actually lands.

Education

  • UC San Diego
    B.S. Bioengineering Systems
    Minor in Business
  • UCLA
    Data Science Certificate
    Fall 2019

Beyond Work

Biohacking, fantasy sports, and South Indian home cooking. I also run a small hobby project for recipes and food storytelling — Rasam Roots.

BiohackingFantasy SportsHome CookingRasam Roots