Senior AI Engineer

QuinStreet

23h•$80 - $100•Remote

About The Position

QuinStreet is a pioneer in powering decentralized online marketplaces that match searchers and “research and compare” consumers with brands. We run these virtual- and private-label marketplaces in one of the nation’s largest media networks. Our industry leading segmentation and AI-driven matching technologies help consumers find better solutions and brands faster. They allow brands to target and reach in-market customer prospects with pinpoint segment-by-segment accuracy, and to pay only for performance results. Our campaign-results-driven matching decision engines and optimization algorithms are built from over 20 years and billions of dollars of online media experience. We believe in: The direct measurability of digital media. Performance marketing. (We pioneered it.) The advantages of technology. We bring all this together to deliver truly great results for consumers and brands in the world’s biggest channel.

Requirements

Production backend Python experience, including async patterns, type hints, packaging, and testing.
Direct production experience designing and configuring AWS ECS/Fargate, SQS, API Gateway, RDS (PostgreSQL), S3, IAM, and CloudWatch, with infrastructure as code (Terraform or CDK).
Real shipped systems calling the Anthropic Claude API in production, with demonstrated experience in prompt design, structured output, error handling, and cost trade-offs.
Demonstrated track record reducing token spend on production LLM workloads, with specific before/after results you can walk through.
Working knowledge of other LLM providers sufficient to recommend cheaper or better alternatives for specific tasks.
Production Playwright experience at scale, including headless Chromium failure modes, network idle detection, dynamic content handling, viewport switching, and screenshot strategy. Selenium or Puppeteer experience does not substitute.
Machine learning fundamentals sufficient to evaluate when LLM analysis is the right tool versus a classifier or rules-based approach, and to reason about evaluation and false-positive rates.
Docker and containerization experience, including image optimization, ECR, and stateless worker design.
Ability to operate fully independently — no engineering team underneath you — while documenting continuously and coordinating cleanly with an internal team.

Nice To Haves

Experience with API ingestion and field-level diff-detection systems.
Laravel or PHP familiarity, enough to coordinate cleanly on API contracts with the portal team.
SOC 2 Type II compliance experience.
Salesforce API integration experience.
Regulated-industry experience (financial services, healthcare, or insurance).

Responsibilities

Design and configure the production AWS environment (ECS/Fargate, SQS, API Gateway, RDS PostgreSQL, S3, IAM, CloudWatch) using infrastructure as code (Terraform or CDK).
Build a stateless, containerized worker fleet that integrates Playwright-based page rendering, structured rule evaluation, and Claude API analysis.
Implement token optimization strategies across the LLM pipeline — prompt engineering, context pruning, caching, model selection, and batching — with measurable cost outcomes.
Define and document API contracts, job payload schemas, and database write patterns with the internal Laravel portal team to enable parallel development.
Build third-party API ingestion and field-level diff-detection logic that automatically adjusts monitoring rules when product data changes.
Handle modern web rendering challenges at scale: JavaScript-heavy SPAs, interstitials, cookie consent overlays, dynamic content, viewport switching, and full-page screenshot capture.
Evaluate when LLM analysis is the correct tool versus a classifier or rules-based approach, and design the two-stage rule-engine-plus-AI pipeline accordingly.
Build and maintain a unit test suite covering all modules and APIs to ensure uptime and proper functionality.
Document every architecture decision, configuration, API contract, and operational procedure continuously — not as a final-week deliverable.
Deliver a complete runbook and knowledge transfer to the internal team at engagement close.
Operate independently end-to-end while coordinating closely with the internal portal team and reporting directly to the Senior Director, surfacing risks and trade-offs early.