Open to opportunities

Jiamin Cai

@jiamincai

Message

Backend-focused AI engineer building reliable, production-grade LLM systems and developer platforms.

United States

Message

What I'm looking for

I seek a role building production AI/ML infrastructure or platform services with strong engineering practices, reliability focus, and opportunities to scale agent/LLM systems.

I am a backend-focused AI engineer and machine learning practitioner who builds reliable, production-grade systems that power LLM agents and developer platforms. I design cloud-native architectures, implement semantic retrieval and evaluation pipelines, and ship safety and billing guardrails for end-to-end AI developer experiences.

My work spans production GCP backends, multi-provider agent runtimes, and scalable ML pipelines that improved inference latency and predictive performance; I contributed open-source tooling adopted by hundreds and published research on algorithmic game-theoretic pricing and optimization.

Experience

Work history, roles, and key accomplishments

Current

Founding AI Engineer

Current

Prompt Driven, Inc.

Sep 2025 - Present (9 months)

Led backend and GitHub App development for an end-to-end AI developer platform, scaling OSS adoption (400+ stars) and enabling 100+ developers to use the GitHub App; improved one-shot success by 20% via semantic few-shot retrieval and operated a production GCP backend for reliable async jobs.

GCP Pub Sub Firestore Go Python (Backend)Vector Search Distributed Systems Reliability Engineering

Machine Learning Scientist (Intern)

TWG Global Holding

Jun 2025 - Aug 2025 (2 months)

Built cloud batch company-scoring pipelines using Palantir AIP agents and a Java multithreaded inference pipeline, reducing end-to-end inference time by 66% and improving predictive factors with Instrumented PCA to triple out-of-sample R².

Java Multithreading LightGBM Model Ensembling Cloud Infrastructure

Reinforcement Learning Research Assistant

The Chinese University of Hong Kong

May 2023 - Oct 2023 (5 months)

Designed simulation experiments to detect collusive algorithmic pricing, adapted evolutionary game-theoretic multi-agent Q-learning showing ε-approximate NE convergence, and refactored code to multithreaded C++ achieving up to 12× speedup.

Reinforcement Learning Multi Agent Q Learning C++Simulation Performance Optimization Python

Education

Degrees, certifications, and relevant coursework

Carnegie Mellon University

Master of Science, Information Systems Management

2024 - 2025

Grade: GPA: 3.91/4.3

Activities and societies: Teaching assistant for Distributed Systems; coursework-focused projects in cloud security and applied ML.

M.S. in Information Systems Management with coursework in Distributed Systems, Database Management, Cloud Security, Applied Machine Learning, and Unstructured Data Analysis; Dean’s Honor List (GPA: 3.91/4.3).

The Chinese University of Hong Kong

Bachelor of Engineering, Financial Technology

2020 - 2024

Grade: First Class Honors (Major GPA: 3.8/4.0)

Activities and societies: Undergraduate research assistant in reinforcement learning and multi-agent systems; relevant project and publication work.

Bachelor of Engineering in Financial Technology with First Class Honors and a Major GPA of 3.8/4.0; coursework included Systems Programming, Data Structures, Cyber Security, E-payment Systems, Optimization Methods, and Web Development.