Live, LLM-powered systems augmenting thousands of contact-center agents at Capital One.
A public rebuild of the shape of the systems below, running live in your browser. Talk into your mic — or play a sample call — and watch a two-voice transcript, self-drafting notes, RAG-retrieved procedure docs, and a sub-second frustration alert. Deepgram + OpenAI composed into a real-time product; no proprietary anything.
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.
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.
An LLM system that detects and categorizes customer complaints on live transcripts and escalates instantly — replacing a legacy ML classifier.
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.
Models that decision billions of dollars in auto lending every year.
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.
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.
Where I sharpened the fundamentals — research, big data, and NLP.
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.
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.
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.
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.
Want the full detail, with dates and stack?
View my résumé