Climate risk researcher — spatial modeling & public decision support

• Designed and ran a pre-specified benchmark of four flood-hazard layers (FEMA Effective SFHA, FEMA National Flood Hazard Layer, and two NJDEP climate-adjusted flood elevation scenarios) across 671,494 parcels and $312.7B of assessed value in four contrasting counties, drawn from New Jersey's 3.46 million parcel tax composite.
• Quantified disagreement with confusion matrices, Cohen's kappa, and value-weighted metrics: in tidal counties the state layer flags 19 to 24 percentage points more parcels at risk than FEMA's effective layer, worth $11.3B of decision-flipping value in Atlantic County alone, while 7% to 34% of parcels flagged by one FEMA product go unflagged by the other.
• Established layer choice as a policy variable with measurable dollar consequences for flood-insurance enforcement, permitting, and municipal tax planning: fiscal-exposure rankings proved layer-fragile, with Kendall's tau as low as 0.191 between two defensible layers.
• Wrote the limitations into the paper rather than around them: commercial risk models stay outside the benchmark pending license, pilot results stay exploratory until OSF pre-registration is filed, and Sandy and Ida ground-truth validation is scoped, not implied.

• Built an open-source pipeline (Llama 3.1 70B, PySAL, GeoPandas) in which a language model proposes cartographic intent while a deterministic engine owns every statistical and geometric decision across six validation gates, each recorded in a machine-readable execution trace for reproducibility.
• Validated on 530 Atlanta census tracts joining ACS income and CDC PLACES asthma data; the system surfaced a significant income and health spatial cross-correlation it was never asked to look for, and corrected a skewed classification that lifted goodness-of-variance-fit from 0.77 to 0.86.

• Analyzing interagency coordination, institutional barriers, and the gap between a funded municipal climate-adaptation program's design and what it delivers.

• Built a statewide Python pipeline (DuckDB, GeoPandas, tippecanoe and PMTiles) scoring all 3,478,722 New Jersey parcels across 21 counties and 564 municipalities against current FEMA flood zones, NJDEP sea-level-rise layers, and tract-level NFIP claims, finding 19.31% of parcels and 25.86% of value exposed.
• Held the work to a published quality contract: 6 of 7 automated gates pass and the seventh, an 88.34% tax-record join rate, ships as a disclosed limitation rather than a quietly relaxed threshold. The heaviest stage runs in 1.89 hours against a 12-hour budget and is county-resumable throughout.
• Extended the same parcel core into a five-scenario planning dashboard (current flood, sea-level rise, Category 1 and 3 surge) scoring exposure, sensitivity, and consequence against CDC Social Vulnerability Index and NJDEP overburdened-community context; pipeline and statewide QA complete, deployment pending.

• Ingested Rutgers University's statewide coastal inundation product for 8 New Jersey municipalities across 21 water levels (0 to 20 ft above mean higher high water) to report which critical facilities and roads are exposed at each level, with per-town operations reports and CSV export.
• Kept the seam between the published hazard model and this project's own inference explicit: per-facility depth is computed only where a tidal datum offset can be verified (7 of 8 towns), and the methods page publishes the depth error budget and the rate at which the two models disagree.

• Conducted build-out analyses for 7 municipalities under four flood inundation scenarios (FEMA 1%, FEMA 1% plus 3 ft, 2 ft, and 5 ft) supporting Climate Change Hazard Vulnerability Assessment reports, one of which Pennsville Township formally adopted and used to propose amendments to local land use and zoning policy; automating the workflow in Python (ArcPy) and ModelBuilder raised throughput 60%.
• Designed pipelines digitizing Hazard Mitigation Plans across all 21 New Jersey counties (ArcPy, NJ Geocoder, FME ETL), plus a Survey123 action tracker whose automated text extraction and validation cut manual data-entry errors 95% and raised geocoding accuracy to 100%.
• Authored the methodologies, metadata, and validation procedures behind these datasets to Center and University publication standards, and released the results as public web applications and dashboards municipal staff use directly.

• Applied LEHD employment data to measure how post-pandemic remote work reshaped daytime population and worker distribution in Philadelphia's Chinatown, shaping business-district and equitable-development recommendations.
• Modeled neighborhood demographics with Census data for an Equitable Development Plan built on input from 70 or more residents across 6 sessions.
• Designed the surveys and focus groups behind that input, and presented spatial findings to residents and municipal decision-makers in non-technical terms.

• Engineered automated data pipelines (Python, Census APIs) that cut processing time 60%, enabling faster analytical turnaround across the state.
• Produced housing affordability findings covering 12 New Jersey cities for a multi-city research publication, presented to municipal planning officials.
• Mapped commuter patterns from LEHD origin-destination data and co-developed policy frameworks with faculty to support state and local decisions on housing equity.

• Co-authored a feasibility study finding a 42% net-present-value advantage for blue-green over gray infrastructure across 5-year and 25-year storms; it informed $12M in state investment and was cited in national flood-resilience policy debate.
• Screened 1,000 candidate open spaces with suitability models weighting ecological, social, and financial criteria, and led a city-scale resilience strategy that helped secure $9M from India's National Disaster Management Authority.
• Wrote a Sponge Parks handbook adopted by the Greater Chennai Corporation as a technical and policy reference, and ran capacity-building workshops for 30 or more municipal engineers and department heads.

Rutgers University, Bloustein School of Planning and Public Policy · Expected Dec 2026

• Dual Master of City and Regional Planning / Master of Public Informatics (MCRP-MPI)
• Graduate Certificate in Geospatial Information Science and Technology (in progress)

Anna University, School of Architecture and Planning, Chennai, India · May 2021

• Bachelor of Architecture, First Class

Columbia University and Northeast Big Data Innovation Hub · 2025

• Transportation Data Science Program, National Student Data Corps (NSDC)

Council of Architecture, India

• Licensed Architect (CA/2023/158480)