Adam Epstein San Francisco · open to new roles

Adam Epstein

Data Scientist and Machine Learning Engineer with four years building production ML and AI systems across security, sales intelligence, and analytics. Most recently I built LLM guardrails, a detection engine, and a risk-scoring pipeline at Vezran.

Explore my work Download resume Get in touch
4+ yrsProduction ML and AI
MSData Science, UOP
01

About

I'm a data scientist and machine learning engineer based in San Francisco. My work sits in the layer between a model and everything it shouldn't do: screening prompts before they reach an LLM, scoring how much to trust an account, and catching the attack that looks ordinary one event at a time.

Before security, I spent time on the commercial side — lead scoring, deal prediction, and federated learning for a sales intelligence product — and before that on analytics and RAG pipelines. The through-line is the same: take a messy signal, make it into a number someone can act on, and make it hold up in production.

I studied Data Science at the University of San Francisco and then at the University of the Pacific. Right now I'm looking for Data Scientist, Machine Learning, and AI Engineering roles.

02

Experience

Five roles across security, AI products, real estate analytics, and aviation.

Dec 2025 — Jul 2026

Senior Machine Learning Engineer

Vezran · Remote
  • Built a three-part defense-in-depth platform: a risk-scoring pipeline, a polyglot detection engine, and an LLM guardrail, wired into a closed loop where trust scores tune detection sensitivity.
  • Ran 10+ models into a single 0–100 trust score driving four tiers of automated response.
  • Shipped the guardrail as a FastAPI microservice on AWS Lambda with injection detection, a 30-pattern OWASP LLM Top 10 library, a semantic check, and six-type PII redaction.
  • Hardened it with circuit breakers, quorum consensus, and SHA-256 hash-chained attestation, validated by 430+ tests in GitHub Actions.
PythonRustFastAPIAWS LambdaTerraformDynamoDB
May 2025 — Dec 2025

Senior AI Engineer

Otom.ai · Palo Alto, CA
  • Built an AI-native lead scoring system combining engagement signals, stakeholder depth, ICP fit, deal momentum, and buyer intent into one normalized score.
  • Built a deal-nudging system that detects stalled opportunities and drafts three-sentence follow-ups from CRM notes, email threads, and meeting transcripts.
  • Shipped a federated learning (FedAvg) pipeline training per-user models to predict which opportunities close within 45 days.
PyTorchMongoDBNLPFedAvg
Jan 2024 — Jul 2025

Senior Data Analyst

City Real Estate · San Francisco, CA
  • Turned raw transaction data into graphs used daily by 55 agents and their clients.
  • Designed the quarterly listings-and-sales newsletter, analysis and visuals included, for the full client list.
SQLTableauPython
Jul 2023 — Sep 2023

Software Engineer

Supernal · Fremont, CA
  • Automated weather data collection through API calls and produced seasonal trend reports for flight operations forecasting.
  • Built a model that gave safety teams a head start on analysis previously done by hand.
PythonAPIsForecasting
Jun 2022 — Aug 2022

Data Scientist

Jones Lang LaSalle · San Francisco, CA
  • Built a text validation framework converting unstructured data into structured arrays for the engineering team.
  • Developed binary classification models that sharpened customer targeting.
  • Built end-to-end RAG pipelines in LangChain: preprocessing, embeddings, vector storage, and LLM integration.
LangChainRAGClassification
03

Projects

Systems I owned end to end. Open one for the full story.

04

Skills

Languages

Python for everything, Rust where latency matters, SQL for the parts people actually query.

PythonRustSQLJavaC++R

ML & AI

Supervised models, federated training, and LLM systems — RAG pipelines, prompt engineering, and guardrails.

PyTorchLangChainRAGNLPFedAvg

Deployment

Services that survive contact with production: typed APIs, infrastructure as code, tests in CI.

FastAPIAWS LambdaTerraformKubernetesGitHub Actions

Data & visualization

Getting the data honest before anyone models it, then making the result readable by non-engineers.

MongoDBDynamoDBTableauPandas
05

Education

University of the PacificMS, Data Science · San Francisco, CA 2023
University of San FranciscoBS, Data Science · San Francisco, CA 2020
06

Hobbies

What I get up to when I'm not in front of a terminal.

Golf
Pickleball
Movies
Friends
Music
New restaurants
07

Reach out

adamep1@gmail.com 415-705-9227 San Francisco, California
Resume PDF GitHub