Johnson Controls / Qolsys Inc. 

• Led a California-based team of data scientists and engineers, managing end-to-end data pipeline and analytics projects and MLOps deployment
  

• Bridged day-to-day operations with leadership team's strategic goals; Directed multifaceted projects, from concept to deployment, in areas such as product recommendation

• Built 5-class audio classifier with 92% accuracy using transfer learning based on TensorFlow's YAMNet; integrated with Silabs microcontroller to receive streaming audio signals
  

• Built remaining-use-of-life predictive ML model for PowerG+ device batteries using Monte Carlo simulation, survival analysis, random survival forests, clustering / classification. Deployed POC to Flask app/RESTful API. Collaborated with hardware team to collect data from in-lab experiments

• Advisor for GenAI Q&A chatbot project, increasing user engagement via interactive capabilities with products; Designed feedback function and data model, translating designs to deployment on Firestore with Full Stack team

• Building NoSQL graph database (Neo4J) to explore dynamic relationships of alarms and sensors, creating context to send to central station. Triangle counting to identify collaborative alarm response, tight-knit sensor networks, and detect unusual patterns

• Initiated collaborations across Hyderabad and Toronto's device level/processing layer teams to standardize data models and IoT event structures, enhancing new and existing product support and event handling

• Designed and built internal data catalog tool with JSON data from GCP storage, optimizing 615 GB+ daily data ingestion and engineering efficiency by 88%
  

• Deployed edge-based computer vision model on panel using TensorFlow/TFLite
  

• Finalist for Tech Challenge (2022), presenting Geofence project + Patent submission
  

• Analyzed, designed and visualized historical data and behavioral events in PowerBI; published reports monitoring KPIs, used by client-facing teams
  

• Led the development of a scalable data platform, enhancing data quality and accessibility for analytics and BI functions 
  

F-Secure

• Designed and created dashboard with new KPIs for cost reduction using PostgreSQL in Redshift
  

• Performed cluster analysis to understand customer behavior
  

• Updated and maintained existing SQL queries and dashboards on Redash
  

Statistics Without Borders

• Collaborated with The Cheetah Conservation Fund zoologists and researchers to analyze 24 years of questionnaire data; Used mixed-effects statistical models, normality/correlation tests in R on longitudinal data; interpreted results of multiple imputations used on missing data to inform conflict resolution strategies
  

EDUCATION & CERTIFICATIONS

Graduate Certificate in Theoretical Statistics and Probability

M.Sc. in Data Science

B.Sc. in Cognitive Science w/ Specialization in Computation

PUBLICATIONS

PROJECTS

Tech Communities and Data Job Market Alignment: A Cross-Platform Analysis

• 2024 dbt Modeling Challenge with Social Media Data w/ partnership between Paradime, MotherDuck, Hex

Improving Product Order Relevancy for E-Commerce

• Collaborative coursework with partner company K-Rauta
  

• Produced search engine demo on Flask that improved product order relevancy upon user search with Multinomial Naive Bayes algorithm, calling data with REST API Authentication

Visualizing Earthquakes with data called from USGS's API

• Used D3.js to create histogram & bar graph to visualize earthquake magnitudes
  

• Created a heat map of earthquakes using Google Maps Javascript API
  

• Built web app with Flask and deployed onto Heroku

UNDERGRADUATE RESEARCH

• Paper accepted to the Design Automation Conference for Poster Session (2017)
  

• Presented project at the 29th Annual Undergraduate Research Conference at UC San Diego

• Assisted in optimizing feature matching in the SFM algorithm using Intel Intrinsics (Similar algorithm: SFM Toy Library)
  

• Presented project at the 28th Annual Undergraduate Research Conference at UC San Diego