Applied AI · Salt Lake City

Michael Northrup

Forward-deployed AI engineer and tech lead. I work directly with business teams to find the problems worth solving, then build and ship the production AI systems that solve them, plus the playbooks that let other teams do the same.

Tech leadand one of the lead architects for AI Ops at BambooHR
Agents & evalsshipped for payroll anomaly detection, onboarding audits, and customer signals
8+ yrsshipping ML and AI, including models that contributed to a $3M/month revenue lift
MN

Ask about Michael

An AI assistant that knows my background

Live

Hi! I'm an AI assistant that knows Michael's background. Ask me about his experience, what he's built, how he works, or his skills.

Fit check: paste the full job description and press Send.

AI-generated answers can be imperfect; for anything important, email Michael. Questions are logged anonymously to improve answers.

Selected work

BambooHR · Payroll · Agents and evals

An AI auditor that payroll experts can actually check

Passed go/no-go review · method reused for a second audit product

Problem
Payroll mistakes are expensive and experts can't review everything. An AI that flags anomalies is useless if the experts can't see why.
My role
Wrote the spec, built the system, and worked it through rounds of feedback with payroll specialists.
Approach
An agent-based anomaly detector with evals from day one and human review of every finding. Built a lineage tool that lets experts verify each detection rule down to its source data, with sign-offs reset whenever a rule changes. Reused the same method for document AI that reads a new client's tax filings, checks them against their payroll setup, and links every finding to its source.
BambooHR · Platform · Architecture

Making the next AI app a folder, not weeks of tickets

Platform owner · first version live in staging, first apps moving on

Problem
Every internal AI app rebuilt the same sign-in, access control, data and PII access, LLM access, and deployment, and waited weeks on the same approvals.
My role
Owner and architect. Audited the existing apps, wrote the plan and architecture decisions, ran the cross-team design session, and shipped the first version.
Approach
Pay for identity, governed data and PII access, LLM access, and deployment once at the platform level, with fail-closed access by default. A new app becomes a folder and a config file.
BILL · Fintech · Customer experience

Turning an ignored decline problem into $3M a month

$25M payment-volume opportunity · $3M/month revenue increase

Problem
Failed customer payments were costing revenue and hurting the payment experience, and the problem had gotten little attention.
My role
Found it, built the business case, pitched it, and drove it end to end: models, interventions, and the production ML foundation underneath.
Approach
Predicted half of transaction declines ahead of time so the business could intervene. Built it on a monitored SageMaker and S3 data lake with drift detection, plus a Feature/Train/Inference/Production framework that made retraining repeatable company-wide.

Experience

Feb 2026 – now

Sr. AI Solutions Engineer & Tech Lead

BambooHR · Salt Lake City

Internal forward-deployed AI engineer for CX, payroll, revenue, and operations teams, owning projects from discovery through deployment, adoption, and handoff. Tech lead across AI Ops projects and one of the team's lead architects since October 2026.

  • Built an agent-based payroll anomaly-detection system with evals from day one, plus a lineage tool that lets payroll experts verify every rule against its source data.
  • Shipped document AI that extracts data from scanned, multi-format tax filings, checks it against a new client's payroll setup, and links every finding to its source.
  • Grew a customer-interaction intelligence pipeline from sentiment classification into product, service, and sales-expansion signals across phone, chat, support, and meetings.
  • Own an internal AI application platform that handles sign-in, access control, data and PII access, LLM access, and deployment once, so a new app is a folder and a config file instead of weeks of tickets.
  • Built a coding-agent curriculum for the technology org, launched an AI Champions program, and run workshops, hackathons, and leadership AI training.
Jan 2025 – Jan 2026

Data Scientist II

BILL · Draper, UT
  • Found an overlooked transaction-decline problem, built the business case, and drove it from pitch to production, surfacing a $25M total-payment-volume opportunity by predicting 50% of declines.
  • Deployed decline-intervention models that contributed to a $3M monthly revenue increase and a better customer payment experience.
  • Architected the production ML foundation (SageMaker, S3 data lake, monitoring, drift detection) and a company-wide Feature/Train/Inference/Production framework.
Nov 2023 – May 2024

Senior Data Scientist

Preston Ventures · Aliso Viejo, CA
  • Engineered ETL and labeling pipelines and deployed validated healthcare models, with MLflow tracking for reproducibility across teams.
Nov 2022 – Nov 2023

Full Stack Data Scientist

Bayesian Health · New York, NY
  • Built and deployed clinical decision-support models on EHR data into customer-facing apps, with state-machine decision logic and MLOps foundations (registries, monitoring, automated deploys).
Aug 2018 – Feb 2022

Associate Data Scientist

BioFire Diagnostics · Salt Lake City
  • Built a role-based, HIPAA-compliant data app, automated validation for customer-facing pipelines, and modernized analysis infrastructure on AWS.
2017 – 2018

Earlier

Intermountain Healthcare · RMCOEH
  • Survival-analysis models for turnover risk and applied occupational-health research.

How I work

Skills

Applied AI
AI agents, LLM applications, RAG, LLM evaluation, MCP, human-in-the-loop workflows, multi-tier classifiers
ML
Predictive modeling, anomaly detection, survival analysis, forecasting, A/B testing, MLOps
Engineering
Python, TypeScript, SQL, R · FastAPI · Docker, Kubernetes · AWS · Databricks, Snowflake, dbt, MLflow, Spark · Okta/OIDC · OpenTelemetry
Leadership
Technical leadership and architecture, AI enablement and training, cross-functional delivery, business cases
Domains
Customer experience, HR tech, fintech, healthcare

Education

Master of Statistics

University of Utah

B.S. Mathematics & Statistics

University of Utah