Allah-u-AbhaRodrigues

Selected as a Governance, Rule of Law and Peacebuilding (GRP) Fellow supporting UNDP's work on monitoring, evaluation, learning, communications, and peacebuilding/local development. Working with the GRP Hub to help strengthen how governance and rule-of-law programs measure impact, communicate results, and translate field evidence into actionable learning across global and country-level initiatives.

• Supported Monitoring, Evaluation & Learning work for governance, rule of law, and peacebuilding initiatives across UNDP's GRP Hub
• Contributed to research and synthesis around peacebuilding, local development, civic space, and people-centered justice/security programming
• Helped structure feedback and analysis for impact-measurement tools, including indicator frameworks and approaches to evaluating program effectiveness
• Applied a technical background in computer science, statistics, and data systems to improve how development programs organize evidence, measure outcomes, and communicate results
• Worked across international, cross-functional teams spanning policy, communications, reporting, operations, and program management

Building Regilon, a property-tax intelligence platform for commercial real estate portfolios. Uses machine learning, statistical modeling, and document intelligence to detect overassessments, estimate potential savings, and surface evidence before appeal deadlines. Focused on property taxes — one of the largest recurring CRE expenses — where missed deadlines and inaccurate assessments quietly erode NOI and asset value.

• Built ML pipeline combining county assessor rolls, CoStar, MLS, and Regrid data to score overassessment likelihood across a portfolio
• Designed deterministic comparable-sale engine: opportunity scores derived from comp strength, deadline urgency, and data completeness
• Shipped automated deadline tracker monitoring appeal windows across CA, TX, NY, and CT with supplemental-assessment flagging
• Architected evidence-packet generation workflow: source-cited PDFs ready for county appeal boards in 2–3 business days

Software engineer on the Yalies.io team, one of Yale's largest student-run platforms serving 30,000+ students and alumni. Expanded backend search and scalability to reliably handle 3,000+ daily requests.

• Expanded backend and search to serve 3,000+ daily requests for 30,000+ users via paginated endpoints, request-level caching, and monitoring (Next.js/Node + Elasticsearch)
• Improved directory and profile experience by shipping query filters, profile views, and contributor docs
• Reduced dev ramp-up time via concise API specs and runbooks

Built a Streamlit-on-AWS analytics engine for a $32M DTC brand, reducing marketing spend by 2% through demand forecasting, Monte Carlo scenarios, and channel-reallocation recommendations.

• Reduced marketing spend by 2% for a $32M DTC brand by building a Streamlit-on-AWS analytics engine with Random Forest demand forecasts and Monte Carlo scenario modeling
• Packaged forecasts, scenario controls, and KPI tracking in a single dashboard with automated data refresh and audit logs
• Improved leadership decision speed by surfacing channel reallocation recommendations with confidence intervals

Scaled a photorealistic 3D real-estate web app from ~200 to 1,000+ users. Delivered sub-200ms property lookups and increased listing-detail clicks by 35%.

• Scaled photorealistic 3D real-estate app from ~200 to 1,000+ users by integrating CesiumJS with Google Photorealistic 3D Tiles in Next.js/TypeScript and adding Street View handoff + click-to-building fly-to
• Delivered sub-200ms property lookups by designing cached geospatial endpoints fusing Zillow CSVs, ATTOM API data, and Microsoft Building Footprints with EPSG:4326 mapping and Haversine search
• Increased listing-detail clicks by 35% and cut time-to-first-insight by 40% by prototyping Interior Transparency using PyTorch + Hugging Face to infer floor layouts and render 3D interior previews
• Reduced release cycle from 2 days to 45 min by building a Docker-first CI/CD pipeline (GitHub Actions → AWS ECS) with 88% unit-test coverage and blue-green deployment

Lead 1:1 and group mentoring sessions for middle and high school students in software development, robotics, Python, and AI fundamentals. Design and teach hands-on projects to make computer science accessible.

• Delivered a full-stack CRUD app in Java/Spring Boot by mentoring a cohort of 4 seniors on OOP design, testing, and CI
• Built an NBA fantasy predictor to AUC ≈ 0.95 with a junior mentee by integrating 5 Kaggle datasets, XGBoost tuning, and a Streamlit app for lineup exploration
• Designed hands-on projects spanning game development, Arduino automation, and Python robotics systems