Machine Learning Engineer

Shields Group Search; San Francisco (CA); On-site; USD 220,000 - 300,000; Full time

Benefits offered by this job

Equity

Onsite work five days a week

Hybrid option

Job summary

Shields Group Search is partnering with a well-funded AI infrastructure startup to build the trust and security layer for AI agents. The opportunity focuses on real-time verification and decisioning infrastructure that distinguishes legitimate agents from harmful activity.

As a Machine Learning Engineer, you will own end-to-end ML systems—from data pipelines to real-time serving—developing models for fraud detection, adversarial ML, and anomaly detection under strict latency constraints.

Qualifications

• 3+ years of professional ML experience with production models.
• Strong Python and a modern ML framework (PyTorch or TensorFlow); solid SQL.
• Production ML systems end to end: feature pipelines, training, evaluation, and low-latency serving.
• Comfort with low-label and cold-start problems, including anomaly detection and weak supervision.
• Data analysis, statistics, and experimental design.
• Strong software engineering; reproducible, tested code.

Responsibilities

• Build and deploy models that identify fraudulent or malicious AI agent activity in real time.
• Develop behavioral and anomaly detection across automated transactions and API activity.
• Design low-latency ML inference to meet strict latency and throughput requirements.
• Tackle cold-start learning and unsupervised approaches with custom evaluation datasets.
• Own end-to-end ML systems from data pipelines through real-time serving.
• Work on adversarial ML and systems at the intersection of ML, AI agents, and security.

Skills

3+ years of ML production experience

Python

PyTorch or TensorFlow

SQL

Production ML systems

Data analysis & statistics

Software engineering

Ownership & initiative

Tools

Docker

Kubernetes

Git

Job description

Machine Learning Engineer | AI Security & Fraud Detection

Compensation: $220,000–$300,000

Experience: 3+ Years, Ideally 5+ Years

About the Opportunity

Shields Group Search is partnering with a well-funded, early-stage AI infrastructure startup building the trust and security layer for the rapidly emerging AI agent economy.

As AI agents increasingly act autonomously on behalf of individuals and businesses, from booking services and accessing APIs to executing financial transactions, companies need infrastructure that can determine which agents are legitimate, who they represent, what actions they're authorized to perform, and whether their behavior presents a security or fraud risk.

Our client is building real-time verification and decisioning infrastructure that enables businesses to safely interact with AI agents at scale.

The platform combines agent identity verification, behavioral intelligence, risk assessment, and security controls to distinguish legitimate automated activity from suspicious or malicious behavior.

We're looking for an exceptional Machine Learning Engineer to build the models that distinguish a trusted AI agent from an attack, in real time and under load.

This is an opportunity to join a small, highly technical founding-stage engineering team and take meaningful ownership of mission-critical ML infrastructure.

What You'll Be Working On

The models behind this platform operate directly on the production request path. They must evaluate activity and support trust decisions in real time, under the same strict latency constraints as the rest of the infrastructure.

There is no offline batch job to hide behind and no second chance if a prediction is too slow or incorrect.

You'll work on problems involving:

• Real-Time Fraud & Threat Detection: Develop ML models that identify potentially fraudulent, malicious, or unauthorized agent activity while minimizing false positives.
• Behavioral & Anomaly Detection: Identify suspicious patterns, anomalous agent interactions, and emerging risks across automated transactions and API activity.
• Low-Latency ML Inference: Build and optimize production model-serving systems capable of making accurate decisions under demanding latency and throughput requirements.
• Cold-Start & Low-Label Learning: Develop effective detection approaches when labeled examples are scarce, including unsupervised learning, weak supervision, and custom evaluation datasets.
• End-to-End ML Ownership: Build, evaluate, deploy, monitor, and continuously improve production ML systems, from data and feature pipelines through real-time serving.
• Adversarial ML: Help build systems that remain effective as malicious actors adapt their behavior to evade detection.
• Agentic AI Infrastructure: Work at the intersection of modern ML, AI agents, tool calling, identity verification, and security.

What We're Looking For

Must-Haves

• 3+ years of professional ML experience, ideally 5+ years, building, evaluating, and deploying production models, not just research or notebooks.
• Strong Python and a modern ML framework (PyTorch or TensorFlow); solid SQL.
• Production ML systems end to end: feature pipelines, training, evaluation, and low-latency, real-time serving.
• Comfort with low-label and cold-start problems, including anomaly detection, unsupervised methods, weak supervision, and building evaluation sets and labels from scratch.
• Data analysis, statistics, and experiment design.
• Strong software engineering. You ship versioned, testable, reproducible code, not throwaway models.
• Experience working in highly technical engineering environments with strong software engineering standards.
• Ability to take ownership, solve ambiguous problems, and ship production-grade solutions in a fast-moving startup environment.

Strongly Preferred

• Applied ML in fraud, risk, abuse, account integrity, or security.
• Experience building ML models for fraud prevention, threat detection, trust and safety, or other systems designed to prevent malicious activity.
• Graph ML / GNNs, sequence or behavioral modeling, or anomaly detection.
• LLM / NLP work, and familiarity with AI agents, including tool use, function calling, and MCP.
• Experience in adversarial settings, where the threat evolves against the model.
• Explainable, auditable, reproducible ML, with real model versioning.
• Experience developing ML systems at highly technical, engineering-driven companies, particularly those working on large-scale infrastructure, payments, fraud prevention, cybersecurity, or AI platforms.

Bonus Qualifications

• Payment protocols (x402), agent identity (DID / VC), or crypto commitments (Merkle anchoring).
• Experience at an early-stage startup or building ML infrastructure from the ground up.
• Familiarity with real-time transaction monitoring, behavioral risk scoring, or identity verification systems.

Compensation & Work Arrangement

• Equity: Competitive equity participation in an early-stage, venture-backed company.
• Work Arrangement: The team strongly prefers engineers working onsite five days per week. Hybrid flexibility may be considered for exceptional candidates who can regularly collaborate in person.
• Start Date: The team is hiring immediately and can move quickly for the right individual.