Matthew Alhonte, Developer in New York, NY, United States
Matthew is available for hire
Hire Matthew

Matthew Alhonte

Bio

Matt has worked as a machine learning engineer, data engineer, and data scientist for over 10 years, and has worked at the intersection of stats and programming for closer to 20 (before the term data scientist had caught on). He combines strong technical skills in deploying production data pipelines, ML model training, and inference pipelines with a rigorous background in experiment design and statistical inference.

Portfolio

Ophidian Scientific
Amazon Web Services (AWS), PostgreSQL, Keras, XGBoost, Random Forests, Spark...
Syllable AI
Amazon Redshift, Keras, XGBoost, Prefect, DuckDB, Google BigQuery...
Birch Infrastructure
Google Cloud Platform (GCP), BigQuery, Data Build Tool (dbt), Prefect, Python...

Experience

  • Data Science - 15 years
  • Python - 12 years
  • Data Visualization - 11 years
  • Statistics - 11 years
  • SQL - 8 years
  • Machine Learning - 8 years
  • Azure - 5 years
  • Large Language Models (LLMs) - 2 years

Preferred Environment

Git, Jupyter, Visual Studio Code (VS Code)

The most amazing...

...thing I've done is reverse-engineer an undocumented file format containing electrophysiology readings.

Work Experience

Principal AI Systems & Agentic Architecture Consultant

2013 - PRESENT
Ophidian Scientific
  • Designed and provisioned extensible Multi-Agent Evaluation Frameworks and Agentic RAG systems, creating benchmarking pipelines to stress-test autonomous agent tool-use, path routing, and execution safety.
  • Designed and built ETL pipelines in Python, Dask, and Prefect.
  • Oversaw the migrations between Google Sheets and Airtable. Airtable automation was executed in Python.
  • Used operations research libraries in Python to optimize teams for the sports betting website FanDuel.
  • Developed specialized NLP classifiers and evaluators for toxic behavioral detection, pattern classification, and data privacy safeguards across large-scale document corpuses.
  • Built a highly parallelized, fault-tolerant simulation and backtesting framework using JAX to execute adversarial stress-testing and scenario-based model validation.
Technologies: Amazon Web Services (AWS), PostgreSQL, Keras, XGBoost, Random Forests, Spark, Experimental Design, Clojure, Docker, Jupyter, Time Series, Pandas, SQL, Machine Learning, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Operations Research, Data Visualization, ETL, Data Engineering, Data Science, Python, CI/CD Pipelines, Cloud Architecture, Document Parsing, System Architecture, Model Evaluation, Model Monitoring, Kubernetes, Model Deployment, PyTorch, ML Pipelines, Algorithmic Trading, JAX, Biostatistics, Hypothesis Testing, Data Pipelines, Statistics, Airtable, ETL Pipelines, Risk Models, Risk Modeling, Data Scientist, FastAPI, HIPAA Compliance, Node.js, Workflow Automation, Workflow Automation & System Integration, PDF, Architecture, Back-end, Distributed Systems, Automation, Information Design, Healthcare Data Science, Hyperparameter Tuning, Google Cloud Platform (GCP), Model Context Protocol (MCP), WebRTC, Apache Airflow, Twilio, AI Model Training, Forecasting, Probabilistic Modeling, Time Series Forecasting, Claude Code, Time Series Analysis, Software Architecture, AI Integration, API Integration, APIs, Integration, Third-party Integration, Agentic Frameworks, Opus, OpenAI API, Sonnet, Agentic RAG Systems, Multi-agent Systems, Vector Search, Reinforcement Learning from Human Feedback (RLHF), Anthropic, LangChain, Pinecone, Claude API, Weaviate, Health, Healthcare, Healthcare Services, Azure, dbt Cloud, Cloud Platforms, Data Modeling, Data Warehousing, ETL Development, Embedding Models, Machine Learning Operations (MLOps), Data Cleaning, Data Labeling, LangGraph, MLflow, Agentic AI Systems, Pydantic, RAG Architecture, Feature Engineering, Linear Regression, Regression Modeling, AI Chatbots, Chatbots, Artificial Intelligence (AI), Large Language Models (LLMs), AI Design, AI Programming, Agentic AI Systems, Decision Trees, Gradient Boosting, K-means Clustering, Dimensionality Reduction, Ubuntu, Financial Markets

Lead AI Evaluation & Infrastructure Engineer

2021 - 2026
Syllable AI
  • Architected production-grade LLM Evaluation Infrastructure, implementing automated workflows for rapid prompt experimentation, behavioral alignment, and conversation quality metrics.
  • Engineered a high-throughput, asynchronous LLM-as-a-Judge evaluation framework processing hundreds of thousands of multi-step conversational trajectories weekly; optimized parallel execution using structured outputs and batch validation.
  • Deployed and governed 50+ enterprise risk and compliance prediction models, transforming stringent regulatory constraints (HIPAA) into automated pipeline logic serving millions of end-users.
  • Debugged and optimized distributed hyperparameter tuning pipeline, identifying a critical file I/O bottleneck in the Dask cluster that reduced training time from 4 days to 2-6 hours, saving $20,000+ annually in AWS costs.
  • Established CI/CD pipelines for the data science team with Screwdriver; containerized feature engineering pipeline in Prefect, improving deployment reliability; managed Redshift warehouse with dbt.
  • Identified performance bottlenecks in the feature engineering pipeline and rewrote critical sections in Polars, reducing time from 8 hours to 1 hour, and memory footprint from 128GB to 30GB.
Technologies: Amazon Redshift, Keras, XGBoost, Prefect, DuckDB, Google BigQuery, Amazon Elastic Container Service (ECS), Amazon EKS, Terraform, OpenAI, Pandas, Polars, Data Build Tool (dbt), MLflow, PostgreSQL, Azure, Machine Learning, Large Language Models (LLMs), Large Language Model Operations (LLMOps), Machine Learning Operations (MLOps), AIOps, Artificial Intelligence (AI), AI Design, AI Programming, HL7, HL7 FHIR Standard, CI/CD Pipelines, Cloud Architecture, System Architecture, RAG Pipelines, Vector Databases, Model Evaluation, Model Monitoring, Kubernetes, Model Deployment, ML Pipelines, Hypothesis Testing, Retrieval-augmented Generation (RAG), Data Pipelines, Statistics, AI Architecture, AI Voice Agents, AI Tools, ETL Pipelines, Risk Models, Risk Modeling, ETL, Data Scientist, HIPAA Compliance, Workflow Automation, Workflow Automation & System Integration, AI Agents, Architecture, Back-end, Distributed Systems, Automation, Information Design, Prompt Engineering, Healthcare Data Science, Healthcare Software, Agentic AI, Amazon QuickSight, Hyperparameter Tuning, Google Cloud Platform (GCP), HIPAA, AI Agent Orchestration, Text-to-Speech (TTS), AI Model Training, Forecasting, Probabilistic Modeling, Time Series Forecasting, Model Context Protocol (MCP), Claude Code, Time Series Analysis, Software Architecture, AI Integration, API Integration, APIs, Integration, Third-party Integration, Agentic Frameworks, Opus, Slack API, OpenAI API, Sonnet, Agentic RAG Systems, Multi-agent Systems, Vector Search, Anthropic, Claude API, Health, Healthcare, Healthcare Services, dbt Cloud, Cloud Platforms, Data Modeling, Data Warehousing, ETL Development, Embedding Models, Data Cleaning, Data Labeling, Agentic AI Systems, Pydantic, RAG Architecture, Feature Engineering, Linear Regression, Regression Modeling, AI Chatbots, Chatbots, Artificial Intelligence (AI), Large Language Models (LLMs), AI Design, AI Programming, Agentic AI Systems, Decision Trees, Gradient Boosting, K-means Clustering, Dimensionality Reduction, Ubuntu

Data Scientist & Data Architect

2021 - 2021
Birch Infrastructure
  • Assisted with architect data infrastructure for a utility-scale renewable energy and data center company.
  • Created data pipelines with Prefect, mostly stitching together Google Cloud Functions and Cloud Run jobs.
  • Managed BigQuery data warehouse with dbt and made table schemas and transformations.
  • Set up data infrastructure (including Prefect and dbt).
Technologies: Google Cloud Platform (GCP), BigQuery, Data Build Tool (dbt), Prefect, Python, CI/CD Pipelines, Cloud Architecture, Document Parsing, System Architecture, Model Evaluation, Hypothesis Testing, Data Pipelines, Statistics, ETL Pipelines, ETL, Data Scientist, FastAPI, Workflow Automation, Workflow Automation & System Integration, PDF, Architecture, Back-end, Distributed Systems, Automation, Information Design, Time Series Analysis, Software Architecture, API Integration, APIs, Integration, Third-party Integration, dbt Cloud, Cloud Platforms, Data Modeling, Data Warehousing, ETL Development, Data Cleaning, Data Labeling, Feature Engineering, Linear Regression, Regression Modeling, Decision Trees, Dimensionality Reduction, Ubuntu

Senior Data Engineer

2020 - 2021
Endeavor
  • Helped lay the foundations for the customer data platform of one of the world’s biggest media companies, including data pipelines from its subsidiaries into a Snowflake database, and some architecture decisions.
  • Built a data pipeline to do daily replicas of a 15GB database for a subsidiary that managed Super Bowl data, using Prefect and dbt.
  • Developed data pipelines to integrate a subsidiary’s data from its own API, its Mailchimp account, and its Pelcro account.
  • Helped architect a process for productionizing machine learning models with Prefect, dbt, and MLflow.
Technologies: Data Build Tool (dbt), Snowflake, Python, Dask, Prefect, MLflow, Azure, Cloud Architecture, System Architecture, Model Evaluation, Data Pipelines, ETL Pipelines, ETL, Workflow Automation, Architecture, Back-end, Distributed Systems, Automation, Information Design, Amazon QuickSight, API Integration, APIs, Integration, Third-party Integration, dbt Cloud, Cloud Platforms, Data Modeling, Data Warehousing, ETL Development, Data Cleaning, Data Labeling, Feature Engineering, Ubuntu

Senior Data Scientist

2018 - 2019
The University of Colorado — Office of Data Analytics
  • Performed statistical analyses and modeling to support student success and helped establish practices during a restructuring of the university’s office of data analytics.
  • Created and presented findings and visualizations to high-level administrators with Jupyter and Zeppelin.
  • Developed a Monte Carlo simulation-based model to predict semester-by-semester student retention.
  • Built a Bayesian model of re-offense after student misconduct.
  • Modeled the effects of different kinds of financial aid with XGBoost.
  • Created a model to predict student GPAs with scikit-learn and Keras.
  • Helped establish practices during a restructuring of the university’s office of data analytics.
Technologies: Amazon Web Services (AWS), XGBoost, Random Forests, Experimental Design, Data Visualization, Time Series, SQL, Data Science, Machine Learning, Jupyter, Keras, PySpark, Scikit-learn, Pandas, Python, Model Evaluation, Model Monitoring, Hypothesis Testing, Statistics, Risk Modeling, Data Scientist, Information Design, Hyperparameter Tuning, AI Model Training, Forecasting, Probabilistic Modeling, Time Series Forecasting, Apache Spark, Data Modeling, Data Warehousing, Data Cleaning, Data Labeling, Feature Engineering, Linear Regression, Regression Modeling, Decision Trees, Gradient Boosting, Dimensionality Reduction, Ubuntu

Data Engineer

2017 - 2018
NOMI Beauty
  • Designed and built the data infrastructure for a startup that made it easier to book hair-&-makeup appointments in hotel rooms.
  • Architected a big data pipeline with Spark, Kafka, and Cassandra.
  • Built data dashboards in Tableau for the operations team.
  • Designed an ETL for survey data from Typeform's API into MySQL.
  • Created reports in Jupyter notebooks with data visualizations in Python with Altair and Seaborn.
  • Designed and implemented a database schema in MySQL.
  • Designed and supported ETL from Couchbase to MySQL using Python.
Technologies: Amazon Web Services (AWS), Spark, Data Visualization, SQL, Jupyter, Simulations, PySpark, MySQL, Pandas, Python, Data Pipelines, ETL Pipelines, ETL, Workflow Automation, Workflow Automation & System Integration, Architecture, Back-end, Apache Kafka, Apache Spark, Data Modeling, ETL Development, Data Cleaning, Data Labeling, Feature Engineering, Ubuntu

Data Science and Blockchain Integration Consultant

2017 - 2017
Tanktwo, Inc.
  • Architected a blockchain-based solution for managing IoT devices and the data they generate.
  • Create a demo of a potential network using Hyperledger.
  • Simulated a private blockchain network in action, using Python.
  • Helped present a demo to the venture capitalists who were looking to invest.
  • Conducted research on optimal blockchain implementation to suit business needs.
Technologies: Amazon Web Services (AWS), Jupyter, Data Visualization, Time Series, Pandas, Python, Data Scientist, Architecture, Integration, Data Cleaning, Data Labeling, Feature Engineering

Data Science Consultant

2014 - 2017
Hospital for Special Surgery
  • Worked on data science topics in a neurology lab that investigated intraoperative neurophysiological monitoring (IONM)—monitoring muscles and nerves during surgery to prevent damage.
  • Reverse-engineered an undocumented file format containing biosignal data.
  • Attempted to classify nerve conduction readings as indicating injury or anesthesia response using Scikit-learn.
  • Visualized biosignal data with Plotly and presented findings.
  • Investigated using Higuchi Fractal Dimension of nerve conduction readings taken during surgery as a means of assessing potential damage.
  • Analyzed biosignal data with a Python data suite (NumPy, Pandas, and SciPy).
Technologies: Experimental Design, Data Visualization, Time Series, Data Science, Machine Learning, Scikit-learn, Jupyter, Plotly, SciPy, Pandas, NumPy, Python, Biostatistics, Hypothesis Testing, Statistics, Risk Modeling, Data Scientist, HIPAA Compliance, Workflow Automation, Healthcare Data Science, Healthcare Software, HIPAA, Time Series Forecasting, Time Series Analysis, Health, Healthcare, Healthcare Services, Data Cleaning, Data Labeling, Linear Regression, Regression Modeling, Decision Trees, Dimensionality Reduction

Natural Language Processing Consultant

2015 - 2015
New York City Department of Administrative Services
  • Scraped PDFs with Python to help digitize the back catalog for a publication, The City Record.
  • Helped design a schema for entries (such as extracting addresses).
  • Created data cleaning regimens to standardize entries from over a hundred city agencies reported in different formats.
  • Used Python and NLTK to perform exploratory natural language processing (NLP) on a century-long corpus of publications.
  • Worked to integrate a pipeline into their MS Access.
Technologies: Jupyter, Data Visualization, Data Science, Machine Learning, Python, Natural Language Toolkit (NLTK), Document Parsing, Statistics, Optical Character Recognition (OCR), Data Scientist, Workflow Automation, PDF, Data Cleaning, Data Labeling

Integration and Development Consultant

2013 - 2014
Broadband Technologies Group
  • Provided computer vision-based assistance for digitizing video archives.
  • Used OpenCV and Python to tag damaged video areas.
  • Implemented Python to automatically fix certain types of damaged videoes.
  • Helped architect an Android application to deliver simultaneous subtitles for live performances.
  • Prepared presentations with Jupyter.
Technologies: Jupyter, Data Visualization, Python, Data Scientist, Workflow Automation & System Integration, Integration, Data Cleaning, Data Labeling

Research Assistant

2008 - 2013
Hunter College
  • Designed and validated a novel psychometric scale.
  • Analyzed survey data in SPSS.
  • Presented findings at research conferences.
  • Maintained relationships with the lab after graduation, eventually moving from data analysis to Python.
  • Worked on the publication of older data.
Technologies: Experimental Design, Data Visualization, Data Science, SciPy, Python, SPSS, Biostatistics, Hypothesis Testing, Statistics, HIPAA Compliance, Data Cleaning, Data Labeling, Linear Regression, Regression Modeling

Summer Research Assistant

2009 - 2010
Yale School of Medicine
  • Designed and piloted a small study investigating psychopathic traits and behavior during an ultimatum game.
  • Analyzed GSR data.
  • Ran research participants through computer-based tasks in a presentation and DMDX.
  • Analyzed data from surveys and computer-based tasks.
  • Built and maintained a database of participants.
Technologies: Experimental Design, Data Visualization, Data Science, SPSS, Biostatistics, Hypothesis Testing, Statistics, HIPAA Compliance, Healthcare Data Science, Healthcare Software, HIPAA, Time Series Analysis, Health, Healthcare, Healthcare Services, Data Cleaning, Data Labeling, Linear Regression, Regression Modeling

Experience

Making an Ergonomic Interface for Causal Inference (ft Claude Sonnet 4.6)

https://hackersandslackers.com/making-an-ergonomic-interface-for-causal-inference-ft-claude-sonnet-4-6/
A "learning in public" style blog post where I demonstrate using Claude to help me write an easier interface for a causal inference package, specifically pgmpy. I'm picking it up from a book I'm reading on integrating causal inference with ML.

Spring 2018 Complexity Challenge

https://github.com/mattalhonte/sfi-challenge
My entry in the Spring 2018 Complexity Challenge by the Santa Fe Institute.

Graph Theory Notes

This is some code that I wrote to help me to understand the graph theory section of an online course on algorithmic information theory.

Python to Rust

A short walkthrough of training a machine learning model in Python, exporting a model artifact, and serving predictions in Rust. It was accepted as official documentation for a relevant Rust crate called "tract."

Recasting Low-cardinality Columns as Categoricals

https://hackersandslackers.com/recasting-low-cardinality-columns-as-categoricals-2/
A short tutorial on saving memory in Pandas by using categorical variables. It includes a code snippet to take a dataframe and recast the low-cardinality columns as categoricals, saving a bunch of memory.

Removing Duplicate Columns in Pandas

A short tutorial on finding and removing duplicate columns in Pandas.

Downcast Numerical Data Types with Pandas

A short tutorial on saving memory by downcasting Pandas columns into the smallest possible numerical representation.

Sentiment Analysis With AWS SageMaker

https://github.com/mattalhonte/sagemaker-deployment/tree/master/Project
Classifying movie reviews as positive or negative, using SageMaker's version of XGBoost.

Epilepsy Classifier

https://github.com/mattalhonte/epilepsy-classifier
A capstone project for Udacity's machine learning engineer nanodegree.

Splitting Columns With Pandas

I wrote a tutorial on splitting up Pandas columns with nested data.

Education

2006 - 2012

Bachelor of Arts Degree in Psychology

Hunter College - New York City, NY, USA

Certifications

JANUARY 2020 - PRESENT

Machine Learning Engineer Nanodegree

Udacity

Skills

Libraries/APIs

Pandas, OpenAI API, Scikit-learn, SciPy, XGBoost, NumPy, Keras, Dask, Natural Language Toolkit (NLTK), PySpark, TensorFlow, Matplotlib, JAX, PyTorch, Node.js, WebRTC, Slack API, Claude API, Pydantic

Tools

dbt Cloud, Plotly, Jupyter, SPSS, Git, Amazon SageMaker, BigQuery, Prefect, Amazon Elastic Container Service (ECS), Amazon EKS, Terraform, Amazon QuickSight, Apache Airflow, Claude Code, Claude

Languages

Python, SQL, Snowflake, Clojure, Rust

Paradigms

ETL, Functional Programming, HL7 FHIR Standard, HIPAA Compliance, Automation, Model Context Protocol (MCP)

Platforms

Jupyter Notebook, Docker, Amazon Web Services (AWS), Linux, Google Cloud Platform (GCP), Visual Studio Code (VS Code), Azure, Kubernetes, Apache Kafka, Twilio, Ubuntu

Storage

Data Pipelines, PostgreSQL, NoSQL, MySQL

Frameworks

Agentic Frameworks, Spark, Apache Spark, LangGraph

Industry Expertise

Healthcare

Other

Data, Statistical Data Analysis, Exploratory Data Analysis, Unstructured Data Analysis, Complex Data Analysis, Statistical Methods, Statistical Modeling, Statistical Forecasting, Statistical Analysis, Random Forests, Random Forest Regression, Experimental Design, Time Series, Machine Learning, Predictive Modeling, Data Visualization, Data Analysis, Data Analytics, Statistics, Data Science, OpenAI, Artificial Intelligence (AI), Large Language Models (LLMs), ML Pipelines, Model Evaluation, Model Deployment, Model Monitoring, AI Architecture, ETL Pipelines, Data Scientist, Workflow Automation, Workflow Automation & System Integration, Architecture, Prompt Engineering, Hyperparameter Tuning, AI Model Training, Software Architecture, API Integration, APIs, Integration, Multi-agent Systems, Data Cleaning, Data Labeling, Feature Engineering, Decision Trees, Gradient Boosting, Bayesian Statistics, Statistical Programming, Amazon Machine Learning, Tf-idf, Convolutional Neural Networks (CNNs), Analysis of Variance (ANOVA), Dashboards, Data Build Tool (dbt), Deep Learning, Natural Language Processing (NLP), Mathematical Modeling, Data Engineering, Generative Pre-trained Transformers (GPT), Amazon Redshift, Polars, MLflow, Vector Databases, Retrieval-augmented Generation (RAG), Hypothesis Testing, Biostatistics, AI Programming, AI Design, Data Privacy, Machine Learning Operations (MLOps), AIOps, HL7, CI/CD Pipelines, Cloud Architecture, Document Parsing, System Architecture, RAG Pipelines, AI Voice Agents, Airtable, AI Tools, Risk Models, Risk Modeling, AI Agents, Information Design, Healthcare Data Science, Healthcare Software, Agentic AI, AI Agent Orchestration, Text-to-Speech (TTS), Forecasting, Probabilistic Modeling, Time Series Forecasting, Time Series Analysis, AI Integration, Third-party Integration, Agentic RAG Systems, Vector Search, LangChain, Health, Healthcare Services, Cloud Platforms, Data Modeling, Data Warehousing, ETL Development, Embedding Models, Agentic AI Systems, Linear Regression, Regression Modeling, AI Chatbots, Chatbots, Artificial Intelligence (AI), Large Language Models (LLMs), AI Design, AI Programming, Agentic AI Systems, Operations Research, Simulations, DuckDB, Google BigQuery, Pipelines, Warehouses, Algorithmic Trading, Large Language Model Operations (LLMOps), Optical Character Recognition (OCR), FastAPI, PDF, Back-end, Distributed Systems, HIPAA, Opus, Sonnet, Reinforcement Learning from Human Feedback (RLHF), Anthropic, Pinecone, Weaviate, Causal AI, Causal Inference, RAG Architecture, K-means Clustering, Dimensionality Reduction, Financial Markets

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring