Matthew Alhonte
Senior AI Researcher at Toptal
New York, New York, United States
Summary
👤
Senior🎓
Top SchoolMatthew Alhonte is an ML & AI consultant based in New York with 11 years of experience building and productionizing machine learning systems across healthcare, finance, and media. He has a proven track record of optimizing large-scale pipelines—reduced inference and training times dramatically through memory and I/O fixes, rewrote heavy feature engineering in Polars, and saved $20k+ annually by debugging distributed tuning bottlenecks. At Syllable AI he built evaluation infrastructure for LLMs and prototyped a transcript evaluation system that cut costs 50% using OpenAI Batch API, and he currently consults on ML model development and data audits for financial clients. Matthew combines hands-on model deployment skills (TensorFlow→ONNX, Rust inference integrations) with data engineering experience using Prefect, dbt, DuckDB, Snowflake, and Dask. He also shares practical data science insights as a long-running blogger at Hackers and Slackers, reflecting a commitment to clear communication and reproducible workflows. His background in psychology and mathematics gives him an uncommon focus on interpretable, decision-focused models that serve real-world stakeholders.
11 years of coding experience
18 years of employment as a software developer
Bachelor of Arts Major in Psychology, Minor in Mathematics, Bachelor of Arts Major in Psychology, Minor in Mathematics at Hunter College
Stackoverflow
41reputation
3kreached
2answers
2questions
Github Skills (13)
neural-network10
keras10
artificial-intelligence10
model-conversion10
rust10
tensorflow10
onnx10
model-optimization10
python9
apache-spark6
apache-zeppelin6
pyspark6
dask6
Programming languages (8)
Github contributions (5)
sonos/tract
Feb 2020 - Sep 2020
Tiny, no-nonsense, self-contained, Tensorflow and ONNX inference
Role in this project:
ML Engineer
ML EngineerContributions:15 commits, 2 PRs, 14 comments in 7 months
Contributions summary:Matthew contributed to the development of a machine learning model using Keras/Tensorflow. Their work included training a model, converting it to ONNX format, and integrating it with Rust code for inference. The user also focused on shape optimizations within the Rust code to improve performance, showcasing an understanding of model deployment and optimization techniques. The commits demonstrate the end-to-end process of building and deploying a simple machine learning model.
inferenceonnxtensorflowneural-networksartificial-intelligence
mattalhonte/tract
Feb 2020 - Sep 2020
Tiny, no-nonsense, self-contained, Tensorflow and ONNX inference
Contributions:2 PRs, 8 pushes, 3 branches in 7 months
no-nonsenseself-containeddeep-learningtinyinference
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.