Projects

Generative AI & Real-Time NLP

Live, LLM-powered systems augmenting thousands of contact-center agents at Capital One.

Generative AI

Real-Time Call Summarization

An agentic GPT reasoning pipeline on Kafka that auto-drafts servicing-agent notes mid-call, with an LLM "agent-as-a-judge" review stage before anything reaches the agent.

GPT-4 Kafka Python AWS Lambda
−40sAgent handle time
$1.3MAnnual savings
90%Accuracy
Retrieval · RAG

Live Procedure RAG

A retrieval-augmented generation system that runs vector-embedding retrieval over live call transcripts to surface the right procedure and training documents to agents in real time.

RAG pgvector Embeddings Python
80%Document recall
Real timeMid-conversation
Real-Time NLP

LLM Complaint Detection

An LLM system that detects and categorizes customer complaints on live transcripts and escalates instantly — replacing a legacy ML classifier.

LLMs Prompt Eval Python
30% → 80%Detection rate
Real-Time NLP

Frustrated-Customer Alerting

Capital One's first real-time AI alert system — streaming DistilBERT and LLM sentiment over 30k+ live calls a day, alerting managers to de-escalate within about a second of speech.

DistilBERT PyTorch Kafka DynamoDB
~1sAlert latency
30k+Calls / day
200+ TPS@ ~100 ms

Applied ML & Credit Risk

Models that decision billions of dollars in auto lending every year.

Credit ML

Auto-Loan Underwriting ML Suite

Engineered and maintained the machine-learning suite behind Capital One's auto-loan underwriting, finding signal in credit characteristics tied to loan performance to decision billions of dollars annually.

Python XGBoost scikit-learn Snowflake
$32M+Annual value
BillionsDecisioned / yr
Reinforcement Learning

Credit-Policy Simulation Engine

Built a simulation engine that generates synthetic loan applications to power a reinforcement-learning task that optimizes credit policy — testing decisions before they ever touch a real applicant.

Reinforcement Learning Python Simulation
RLPolicy optimization
SyntheticApplication simulation

Earlier Work

Where I sharpened the fundamentals — research, big data, and NLP.

Anomaly Detection · Sprint

Billing-Failure Anomaly Detection

Designed a novel model-training framework for anomaly detection on customer billing data, and built Hive and PySpark pipelines aggregating millions of records. The framework surfaced five distinct classes of billing failure that had previously gone undetected — earning the only extension across an 80+ intern class.

PySpark Hive Python Anomaly Detection
5xDetectable failure types
MillionsRecords aggregated
NLP Research · UPenn

Population-Scale Emotion Classification

Partnered with University of Pennsylvania sociologists on an NLP study — building a supervised emotion-classification model and scraping millions of public social-media profiles to measure population-level emotional shifts after major events. Fed a peer-reviewed sociology research program.

Python scikit-learn NLTK Web Scraping
MillionsProfiles processed
ResearchSociology study
Teaching · General Assembly

Data Science Instruction

Taught a hands-on data-science curriculum on-site at Intuit through General Assembly — walking working professionals from Python fundamentals through machine-learning modeling, and translating dense statistical ideas into intuition they could apply the same day.

Python pandas scikit-learn Jupyter
IntuitEnterprise cohort
End-to-endPython → ML
Deep Learning · Writing

Deep-Learning Projects & Technical Writing

Built and open-sourced deep-learning projects while teaching myself the field — CNN image classifiers, a toxic-comment NLP classifier, and experiments in regularization and dropout — then wrote them up on Medium to make the fundamentals click for other people learning ML.

PyTorch Keras CNNs Medium
Open sourcePublished projects
CNN · NLPDeep learning

Want the full detail, with dates and stack?

View my résumé