Wayne SimmonsOpen to 2027 ML internships
Resume

[00] ML engineer · Queens, NY

I build AI to get things done.

I'm Wayne, an ML engineer in Queens. I build models and tools that find answers in messy data, help people, and take work off their plate.

See the work →[02] About me
[scroll]NYC --:--:--
projects shipped
4
at Apple Retail
4 yrs
CS grad, Queens College
2027
applied to Break Through Tech AI. I got in.
4,300+

[range] Five verbs

What done looks like

  1. [01]find

    Find

    Pull insight out of messy data.

  2. [02]answer

    Answer

    Inform people with grounded answers.

  3. [03]simplify

    Simplify

    Make life easier.

  4. [04]anticipate[next]

    Anticipate

    Act before you have to ask.

  5. [05]delight

    Delight

    Do cool things.

[01] Work

The done list

[all projects →]
[01] Model card[in progress]

CUAD Contract Triage

A pipeline that finds the risky clauses in commercial contracts and ranks which contracts a legal team should open first. Built for Accenture with a 7-person Break Through Tech AI Studio team.

Gets done
Tells a legal team which contracts to open first.
Task
Multi-label clause detection + risk triage
Model
TF-IDF + logistic regression baseline; fine-tuned transformer encoder in progress
Key metric
0.683 macro F1, TF-IDF baseline on the official test split (10 priority clause types)
  • Python
  • scikit-learn
  • Hugging Face Transformers
  • PyTorch
Team project
[02] Model card[done]

What 114,000 Songs Sound Like

An honest look at Spotify's catalog as data, ending in an interactive explorer you can filter yourself.

Gets done
Lets anyone explore 114k songs and test genre stereotypes themselves.
Task
Exploratory analysis + interactive explorer
Model
None. Descriptive stats and correlation analysis
Key metric
~89k unique tracks after deduplication
  • Python
  • pandas
  • Plotly
  • Streamlit
[03] Model card[done]

Decision Tree vs Neural Network

Two models on the Titanic dataset, taken through the full ML life cycle side by side, then scored against Kaggle's real holdout.

Gets done
Shows whether a neural net's extra complexity actually pays off on small tabular data.
Task
Binary classification (passenger survival)
Model
Tuned decision tree vs. 3-layer Keras network
Key metric
0.818 F1 on local test, 0.7608 Kaggle public score
  • Python
  • scikit-learn
  • Keras
  • pandas
[04] Model card[done]

CleanTube

A search-only front door to YouTube. No home feed, no recommendations. Arrive with intent, find it, leave.

Gets done
Lets you use YouTube without getting pulled in.
Task
Cut algorithmic noise out of YouTube
Model
No ML. A single-page HTML wrapper
Key metric
n/a
  • HTML
  • JavaScript
  • Claude

[map] Latent space

How the work clusters

Each dot is a project, placed by what it's about: descriptions embedded with all-MiniLM-L6-v2, then reduced to 2D with PCA. Closer dots, more similar work.

[pc1] →[pc2] ↑Autonomous Data Analyst AgentClaude RAG ChatbotCleanTubeCUAD Contract TriageWhat 114,000 Songs Sound LikeDecision Tree vs Neural Network
  1. Autonomous Data Analyst AgentAn agent that takes a plain-English business question, writes SQL against a real database, runs Python analysis in a sandbox, makes charts, and writes up what it found.[next]
  2. Claude RAG ChatbotA Streamlit chatbot on the Claude API that routes each question to the right knowledge base and answers grounded in it.[done]
  3. CleanTubeA search-only front door to YouTube. No home feed, no recommendations. Arrive with intent, find it, leave.[done]
  4. CUAD Contract TriageA pipeline that finds the risky clauses in commercial contracts and ranks which contracts a legal team should open first. Built for Accenture with a 7-person Break Through Tech AI Studio team.[in progress]
  5. What 114,000 Songs Sound LikeAn honest look at Spotify's catalog as data, ending in an interactive explorer you can filter yourself.[done]
  6. Decision Tree vs Neural NetworkTwo models on the Titanic dataset, taken through the full ML life cycle side by side, then scored against Kaggle's real holdout.[done]

[def] Dictionary

done/dʌn/adjective

1. Finished. Working. Off your plate.

For me: the only status that counts.

[04] Contact

Let's get something done.

Internships, a problem that needs a tool, or just a good playlist. Hire me if you want to get things done.

Email
ElsewhereLinkedIn ↗GitHub ↗
Local time, Queens NY--:--:--