Adeet Patel Email & Phone Number

Charlottesville, Virginia, United States

Open-source GeoMesa contributions: https://github.com/locationtech/geomesa/pulls?q=is%3Apr+author%3Aadeet1- Led a collaborative development team in implementing a Kubernetes controller and CRD using Go(Golang) – ensuring code reliability, exhaustive code coverage, and efficient problem-solving throughreal-time group programming sessions- Created Helm charts for spinning up microservices and deployments more quickly- Managed task tracking through Jira, utilizing DevOps practices to maintain project organizationand accountability- Built an ETL (Extract, Transform, Load) pipeline using Apache Kafka, converting geospatial data tocompatible schema formats and routing it to ingress topics- Optimized Kalman filters / Bayesian statistical models by running the Expectation-Maximization (EM)algorithm on time series data and minimizing stochastic process variance, helping improve air andmaritime tracking software- Implemented unit tests in Java, verifying numerical linear algebra calculations and scientificcomputing algorithms- Designed K6 load tests for a multi-node AWS cluster, identifying and implementing horizontal scalingoptimizations that enhanced throughput by 10x

New York, New York, United States

- Collaborate with Prof. Pascal Wallisch to improve the DS-GA 1001 (Introduction to Data Science) and DS-GA 1004 (Big Data) class curriculum

New York, New York, United States

- Lead a lab section for the course DS-GA 1004 (Big Data), help students debug homework, and hold office hours- Grade homework assignments and provide students with detailed evaluations and feedback

New York, New York, United States

Movie Recommender System Using Spark (Apr 2022 - May 2022)- Trained an alternating least squares (ALS) model on MovieLens ratings, using a Spark cluster- Compared runtime and accuracy of the parallelized ALS model with a LightFM single-machine model- Performed UMAP dimensionality reduction to visualize the learned item representation- Investigated trends and genres of the users who produced the lowest-scoring predictionsModeling Corporate Credit Default Risk (Nov 2022)- Trained an econometric model to predict the probability that a potential corporate borrower will defaulton a prospective loan over the next 12 months- Engineered factors, such as cash flow to debt, debt to equity, etc. with the strongest predictive power

New York, New York, United States

- Trained an agent via curriculum reinforcement learning with an actor-critic framework to play thepuzzle game Baba is You- Provisioned a Google Cloud Platform (GCP) machine on the High-Performance Computing cluster torun algorithms

- Investigated how speakers’ accents (native vs. non-native) or speech impairments (aphasia vs. stutter)triggered errors in the AWS Transcribe and Rev AI automated speech recognition (ASR) systems- Devised a significance testing procedure based on word error rates (WER) computed for each speaker- Deployed an AWS Lambda function for reading the speakers’ audio files from an AWS S3 bucket andrunning AWS Transcribe on each audio file- Implemented a Python script for submitting batches of speech transcription jobs to the Rev AI system

New York, New York, United States

- Optimize code for a named-entity recognition (NER) tool via vectorization, increasing speed by 28%- Write production-level code in Python for deploying the tool onto an Amazon EC2 instance- Design the tool’s front-end web interface using the Flask framework- Build an extract, transform, load (ETL) pipeline on the Civis platform for automating data processing- Take in search keywords as input, scrape the corresponding news/social media data from the Meltwater API, compute metrics from the data via SQL queries, and output a spreadsheet of these metrics

Wilmington, Massachusetts, United States

- Developed in a Linux environment a C++ Qt Quick application that parses XML files of a fixed schema and presents the file contents in a user-friendly visualization, allowing users to view/edit the contents- Created Windows release builds of the application via the MinGW compiler and qmake system- Wrote batch files to fetch and copy .dll and QML dependencies into the build directory- Executed Qt’s Windows deployment tool in order to create the full release build- Created an installer executable via the Qt Installer Framework and InstallSimple- Shipped the release builds to the product specialist and manufacturing teams for their internal use- Enhanced a Python application for mining Raman spectral data from XML files, by allowing the user to whitelist and blacklist keywords from a search of subdirectories within a user-specified directory- Performed statistical analyses on this spectral data- Manually corrected samples misclassified by the Raman spectroscopy instrument

Richmond, Virginia, United States

- Predicted via decision tree classification the direction that Apple’s stock price will move (up/down)- Scraped Google trends data via the pytrends module, and aggregated the hourly data into daily data- Engineered features based on Apple’s historical prices, Wikipedia page hits, and Google News trends- Performed normalization feature scaling, iterative feature selection, and principal component analysis to reduce noise in the time-series dataset and increase model robustness- Created an algorithmic trading strategy based on the machine learning model – generating 31.2% annual return with a Sharpe ratio of 2.8, compared to a buy-and-hold strategy which generated 1% annual return

Charlottesville, Virginia, United States

- Worked on an Agile scrum team to migrate an on-premise workflow system to the AWS Cloud- Led Agile stand-up meetings, ensuring that team members were on track for completing development / quality assurance / etc. for their respective user stories- Tracked cross-team dependencies for user stories and escalated risks and issues where necessary- Collaborated with project managers to create documentation for the newly migrated workflow system- Performed quality assurance testing, devising exhaustive test cases to improve quality- Reported bugs in the newly migrated workflow system front-end and back-end

Charlottesville, Virginia, United States

- Combined multiple ticket sales datasets (one per festival event) via joins to create one universal dataset- Cleaned the universal dataset by fixing spelling mistakes, incomplete data, and double-counting issues- Created various visualizations in Tableau, striving to help improve marketing towards future festivals: (1) a map displaying the number of tickets sold from each state, (2) a pie graph displaying each festival event’s ticket sales as a percent of the total ticket sales, and (3) a time-series graph showing the number of total tickets sold in each month over time

Master Of Science - Ms, Data Science, 3.92 / 4.0

Coursework: Intro to Data Science, Probability & Statistics for Data Science, Machine Learning, Big Data, Advanced Python for Data.

Bachelor Of Arts - Ba, Mathematics, 3.86 / 4.0

Activities and Societies: HackCville (now Forge), Google Developer Student Club (DSC), Code for Charlottesville, Machine Learning Club.

Ib Diploma, 40/45

Education record

Activities and Societies: Mu Alpha Theta, Financial Literacy Club, PBAU High School Mathematics Competitions, Future Business Leaders.