Things I've built
A curated set of projects spanning AI systems, applied products, mobile apps, and ML research.
Everything else
Comprehensive coursework, experiments, and research across ML, statistics, forecasting, and applied analysis.
Payment Method Classifier
Jupyter NotebookArtificial neural network pipeline for payment method classification, with feature engineering and model evaluation.
Neural Sentiment Classifier
Jupyter NotebookMulti-class sentiment classification using recurrent and dense neural network architectures with comparative analysis.
MSFT / Pinterest Valuation
DCF and comparable-company valuation analysis for Microsoft and Pinterest, with scenario modeling and sensitivity tables.
Global Safety Regression
RRegression modeling of global safety indicators using multivariate statistical methods in R, with diagnostic and interpretation analysis.
Multivariate Statistical Methods
Applied multivariate statistics: PCA, cluster analysis, MANOVA, and discriminant analysis across real-world datasets.
The builder
behind the work
I'm a full-stack AI engineer — I work between AI systems and shipped product, taking models and primitives and building the whole app around them.
I came up through industrial engineering, finance, statistics, and product, and each taught me something I use daily. That range is how I end up shipping multi-agent voice systems, calibrated ML for real markets, and iOS apps with on-device vision.
Currently: graduate research in statistical learning at Baruch, hackathon weekends in NYC, and a long queue of product ideas I'm chipping through. Looking for problems where the modeling and the product both matter — and where owning both ends is the unlock.
Currently building
AI-native products
Wins, research,
and where I've trained
AI Blocks
VibeForward × Lovable · Fordham Gabelli, NYC
Recreated Scratch for AI engineering: a visual, node-based way to build AI systems instead of writing everything from scratch.
- 505 adaptive blocks wired into DAG-based system designs
- Two-stage decomposition: pick the stack, then retrieve only relevant blocks via embedding similarity
- Drop in any codebase — AI Blocks detects the stack, links files to blocks, and highlights gaps
- ~80% fewer tokens vs. Opus 4.6 on equivalent tasks
Team
Ryan Rana · Nathanael Lara · Makendy Midouin · Buddhsen Tripathi
Situational Intelligence
Grayscale Hackathon · NYC · Pioneering Minds AI
Real-time AI surveillance system that monitors live feeds, detects emergencies (falls, fights, fires), and understands context before triggering alerts.
- Context-aware reasoning reduces false alarms and enables smarter dispatch
- Pulls city data to prioritize and route responder alerts
- Built end-to-end in 6 hours — shipped with live simulated fall demo
Team
Ryan Rana · Jaiden B · Nathanael Lara
Research Assistant — Prof. Zeda Li
Nov 2025 – PresentResearch Foundation of CUNY · New York, NY
- Designing simulation frameworks for brain connectivity networks
- Implementing network inference methods including Graphical Lasso and Bayesian models
Data Analyst Intern
Sept 2022 – July 2023BCITS Pvt Ltd · Remote
- Large-scale data cleansing across billing and IoT datasets
- Drove 15% improvement in billing accuracy and 20% lift in customer satisfaction
Baruch College — Zicklin School of Business
Expected May 2026MS, Statistics
GPA 3.95 · Regression, Statistical Inference, Multivariate Methods, ML, Data Mining
Imperial College Business School
Sept 2021 – Sept 2022MSc, Investment & Wealth Management
Merit classification · London, UK
Pennsylvania State University
Aug 2016 – May 2020BS, Industrial Engineering
GPA 3.83 · Dean's List, all semesters
Skills & stack
Tools and technologies I use across AI engineering, product, data, and mobile work.
Let's work
together
Open to collaborations, interesting problems, and conversations about AI engineering and products.