Owned delivery of a commercial-insurance data platform spanning 1,000+ property records, turning ambiguous client requirements into product features, success metrics, and roadmaps.
Engineered a cross-field validation framework that surfaced a $40M valuation anomaly, packaged into a decision-ready executive brief that drove client remediation.
Drove AI-assisted validation adoption from 40% to 95% in eight weeks via a catch-log system and field-feedback iteration.
Reduced correction cycles by 20 percentage points working across compliance, actuarial, and data-engineering teams on Jira-tracked sprints.
Product strategyJiraExecutive commsValidation
02 Experiencefiled underProductMarketingAnalyst
Amoga
Dec 2022 – Jun 2023
Product Manager — B2B SaaS
Owned a CRM product end-to-end at a B2B SaaS startup: user research, stories and acceptance criteria, backlog, sprint ceremonies, and the GTM motion around it.
Shipped workflow improvements that cut admin overhead across the sales org.
Drove +7% web traffic and +10% lead conversion via outbound campaigns (Mailchimp, HubSpot) and Superset / GA funnel dashboards.
Built buyer-facing copy, one-pagers, and sales kits; synthesized competitive intel into briefs that shaped roadmap prioritization.
Product ManagementHubSpotApache SupersetGTM
newvibecheck.streamlit.app
03 Projectfiled underSolutions EngProductAnalyst
VibeCheck
2025 – Present
Agentic AI App
An agentic app built end-to-end with Claude Code — and a real production debugging story behind it.
Diagnosed an LLM hallucination failure mode and shipped a YouTube Music API validation layer running 40+ async parallel calls — cut error rate to near zero.
Integrated MCP tool calls, JSON-enforced structured outputs, and session-state memory for multi-turn agentic workflows.
PythonLlama 3.3Groq APIStreamlitMCP
Visit live site
raivana.in
04 Projectfiled underSolutions EngMarketing
Raivana
2024 – Present
Founder — Full-Stack E-Commerce
A live e-commerce platform for authentic Rajasthani handicraft — serverless, multi-currency, and processing real transactions.
Architected a serverless backend with an HMAC-verified webhook handler and idempotency key store (30-day TTL) guaranteeing exactly-once payments.
Built geolocation-based currency routing serving 45+ countries, 8 currencies, and 156 products.
Node.jsNetlify FunctionsStripeVanilla JS
Visit live site
firesight · foundry workshop
05 Projectfiled underAnalystConsultantSolutions Eng
FireSight NYC
Civic · Data
AI inspection prioritization · Palantir Foundry
Consolidates four siloed NYC building-safety databases into one ranked queue for fire-risk inspections — built in the shadow of the 2022 Twin Parks fire that killed 17.
Transparent 0–100 risk score weighting self-closing-door violations, complaint history, and building age across 89,496 Bronx parcels.
On pre-fire data only, the model ranked Twin Parks #1,003 of 89,496 parcels (top 1.1%) — signal the city's siloed systems missed.
Three-view operator workflow with AI-generated dispatch rationales and one-click inspection dispatch.
PythonPalantir FoundryAIP LogicNYC Open Data
Read case study
06 Projectfiled underAnalystConsultant
Retail Stockout Prediction
Data · Team of 4
Stockout risk + inventory optimization
A 51.86% stockout rate, predicted before it happens — then a budget-bounded restocking plan. I led the EDA and visualization that set the strategy.
Led EDA across 5 stores × 8 products — surfaced rainy/cloudy & Saturday peaks and a 'buffer illusion' that broke simple threshold rules.
Team's tuned XGBoost hit 0.77 AUC / 76.6% recall; a Gurobi LP allocated 1,685 units across 11 stores within a $32,836 budget.
PythonXGBoostGurobi
Read case study
07 Projectfiled underAnalystConsultantMarketing
Coupon Acceptance Prediction
Data · Strategy
Coupon acceptance → highway amenity strategy
An NYU analytics project reframed as a planning brief: model which drivers accept coupons, then tell highway planners which amenities to actually build.
Engineered a 57-feature pipeline; tuned Gradient Boosting won at 76.65% accuracy, 80.07% F1, 0.84 AUC.
Turned the model's top predictors into a 'priority amenity' strategy for interstate planners.
PythonGradient BoostingGridSearchCV
Read case study
08 Projectfiled underProductAnalystMarketing
Spotify Product Analytics
Data · Product
NLP for roadmap prioritization
Mined 20,000 real user reviews to prioritize the product roadmap and reduce churn — a PM question answered with data.
Sentiment and theme extraction with NLTK + VADER over 20k reviews.
Translated findings into roadmap priorities and churn-reduction levers.
PythonNLTKVADERPandas
Read case study
09 Projectfiled underAnalyst
RFM Analysis & dbt Analytics
Data
Warehouse-native analytics
Customer segmentation and a modern analytics-engineering stack: RFM in SQL + Tableau, and dbt models on Snowflake.
RFM segmentation on car-sales data in SQL, visualized in Tableau.
dbt Core project on Snowflake with staging + mart models (Jaffle Shop).
SQLdbtSnowflakeTableau
Read case study
§04Toolkit
Fluent from the strategy deck to the production deploy.
01 Product & Strategy
Product Management
Roadmapping
User Stories
Backlog & Sprints
Competitive Analysis
Go-to-Market
Stakeholder Comms
02 Analytics & Data
Python
SQL
XGBoost
Apache Superset
Google Analytics
VADER / NLTK
Funnel Analysis
KPI Reporting
03 AI & Technical
LLM Integration
Agentic Workflows
MCP Tool Use
Structured Outputs
Node.js
REST APIs
Webhooks (HMAC)
Claude Code
04 Tools
Jira
HubSpot
Salesforce
Mailchimp
Git
Streamlit
Excel
PowerPoint
§05About
NYU ’26Yankee Stadium · NYC
I'm a generalist by design, not by accident.
I started in computer science, spent a year in financial-services consulting at Accenture owning a $450M+ data platform, and a stint before that as a product manager at a B2B SaaS startup. Now I'm finishing an MS in Management of Technology at NYU.
The through-line: I take something ambiguous — a vague client ask, a messy dataset, a half-formed product idea — and turn it into something shipped that people can act on. Sometimes that's a strategy deck, sometimes a churn model, sometimes a payments system in production.
I learn fastest by building, which is why most of my projects are live, not slideware. If you're hiring for a role that sits between the technical and the commercial, that's exactly the seam I work in.
Now
MS, Management of Technology — NYU
Before
BE, Computer Science — Ramaiah
Based
New York, NY
§06Contact
Let’s find the seam between the technical and the commercial.
Open to PM, PMM, analyst, consulting, and solutions-engineering roles for 2026. The fastest way to reach me is email.
