Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I am an Applied ML Engineer specializing in building production-grade AI agents and end-to-end automation workflows. I recently architected and deployed a Generative AI system that slashed complex hardware design times from 30 minutes to 1 minute by integrating LLMs with robust Python backends. With deep expertise in multi-agent reinforcement learning (MAPPO) and scalable data pipelines, I am fully equipped to work independently and deliver bespoke, high-impact AI solutions for Twine's diverse clients.…I am an Applied ML Engineer specializing in building production-grade AI agents and end-to-end automation workflows. I recently architected and deployed a Generative AI system that slashed complex hardware design times from 30 minutes to 1 minute by integrating LLMs with robust Python backends. With deep expertise in multi-agent reinforcement learning (MAPPO) and scalable data pipelines, I am fully equipped to work independently and deliver bespoke, high-impact AI solutions for Twine's diverse clients.

Jingxiang Jin

AI Engineer, Data Scientist, Full Stack Developer, +2





I am an Applied ML Engineer specializing in building production-grade AI agents and end-to-end automation workflows. I recently architected and deployed a Generative AI system that slashed complex hardware design times from 30 minutes to 1 minute by integrating LLMs with robust Python backends. With deep expertise in multi-agent reinforcement learning (MAPPO) and scalable data pipelines, I am fully equipped to work independently and deliver bespoke, high-impact AI solutions for Twine's diverse clients.…I am an Applied ML Engineer specializing in building production-grade AI agents and end-to-end automation workflows. I recently architected and deployed a Generative AI system that slashed complex hardware design times from 30 minutes to 1 minute by integrating LLMs with robust Python backends. With deep expertise in multi-agent reinforcement learning (MAPPO) and scalable data pipelines, I am fully equipped to work independently and deliver bespoke, high-impact AI solutions for Twine's diverse clients.

Available to hire

I am an Applied ML Engineer specializing in building production-grade AI agents and end-to-end automation workflows. I recently architected and deployed a Generative AI system that slashed complex hardware design times from 30 minutes to 1 minute by integrating LLMs with robust Python backends. With deep expertise in multi-agent reinforcement learning (MAPPO) and scalable data pipelines, I am fully equipped to work independently and deliver bespoke, high-impact AI solutions for Twine’s diverse clients.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Language

English

Fluent

Work Experience

Operations Systems Coordinator at T&T Property Group

October 1, 2025 - Present

CRM, Routing & Dispatch Support: Maintained 200+ customer records; improved notification workflows (broadcasts + pre-arrival alerts) and planned service routes for nightly dispatch. Service Tickets & Proof Management: Handled 20+ daily tickets; maintained job proof records; resolved customer issues to ensure timely updates and closure.

AI Algorithm Engineer at Miji Tech

September 1, 2024 - Present

PCB Automatic Generation: RL with a shared CNN backbone for PCB placement and 4-layer routing (~200 components), reducing routing time from ~2 days to ~0.5 day. EDA Requirements Assistant: GPT-based EDA requirement tool with prompt engineering and SFT to identify missing/ambiguous requirements and generate client-ready requirement documents. LLM-to-SAT Schematic Generation: LLM-driven pipeline translating requirements into SAT constraints and generating SCH files, reducing creation time from ~30 minutes to ~1 minute per design.

AI Algorithm Intern at Baidu Inc.

September 1, 2023 - January 1, 2024

Search Log Data Quality Pipeline: Spark validation + cleaning pipelines on Hadoop/HDFS over ~1M search queries; defined label-quality standards with annotators and reduced noise from 20% to 6%. Token-Efficient Prompting Pipeline: Built hierarchical prompting for LLM query classification to fit token constraints; validated with prompt-length/few-shot ablations and scaled batch size from 100 to 400.

Algorithm Intern at Miji Tech

June 1, 2023 - September 1, 2023

Auto-PCB Research: Benchmarked Lee, DFS, and A* for Auto PCB routing under different board complexities and obstacle densities, analyzing trade-offs in path optimality and runtime to guide routing strategy choices. Multi-Platform Scraping & Modeling: Implemented an hourly price-collection pipeline across 5 e-commerce platforms; used Random Forest to model feature–price relationships and compare pricing by function. Chip Defect Recognition: Applied data cleaning and augmentation with an ensemble of deep models (VGGNet, CNN, FCN) to classify 100K+ chip defects into 8 types, achieving 99.6% accuracy via model voting.