San Jose, California
Feilian
Huang
AI Software Engineer · Independent Researcher
I build AI systems for engineering and study how machine intelligence changes knowledge, institutions, and technical practice.
Feilian
Professional practice
Work
Engineering experience at the intersection of language models, electronic design automation, retrieval systems, and applied machine learning.
Experience
FEB 2025 · PRESENT
Software Engineer II
Cadence · Full-time · San Jose, California · On-site
- Built PG Agent to automate power-grid routing by translating foundry requirements directly into PSDL with an LLM, reducing more than 1,000 lines of manual add-stripe scripting to near zero.
- Developed ChipRTL, combining LLMs, retrieval-augmented generation, and MCP to automate RTL code generation through high-level synthesis.
- Developed Innostack, a business-unit-wide AI platform that unifies Cadence tools through natural language, with multi-tool support, a VS Code extension, PG Agent embedded through SKILL, and integrations across teams using SKILL and MCP.
MAY 2024 · AUG 2024
Software Developer Intern
Cadence · Full-time · San Jose, California · On-site
- Built a RAG pipeline for the Cadence JedAI LLM. Query decomposition and semantic chunking raised answer correctness on real customer-support questions from 0.46 to 0.71.
- Debugged document ingestion, chunking, hierarchical data organization, and insufficient-context failures in the response pipeline.
- Built a web application for generative-AI customer support using Python and JavaScript.
SEP 2021 · MAY 2023
Research Assistant
Georgia Institute of Technology · Full-time · Atlanta, Georgia
MAY 2022 · AUG 2022
Software Developer Intern
AirMettle, Inc. · Internship · Houston, Texas
- Listed as the fourth author on the patent “Partitioning, Processing, and Protecting Media Data.”
- Designed and implemented automatic segmentation of uploaded videos into clips for parallel processing with video models on AWS, tracking the workflow in Jira in an Agile environment.
- Built a query system for video clips using Python, pandas, and SQL.
- Applied computer-vision models to partitioned video for parallel processing.
- Supported by National Science Foundation Grant 2135007.
DATES NOT SHOWN
Research Assistant
The Johns Hopkins University
Education
M.S. in Computer Science
The Johns Hopkins University
August 2023 to December 2024
B.S., Mathematics & Computer Science
Georgia Institute of Technology
Double major
Publications & active projects
Research
Two distinct research programs: empirical and systems work on AI, and the reconstruction of American maritime activity from 1783 to 1815.
AI & computing
2026Unverified Foundations: How AI-Generated Mathematics Builds on Itself
Data and code archived on Zenodo. arXiv preprint submitted.
The First AI-Saturated Election: Disclosure, Detection, and the Regulatory Limits of Generative Political Persuasion
With Yujia Liu. Submitted to Election Law Journal.
Client-Side Measurement of Free-Tier LLM Inference
Submitted to SIGMETRICS 2027. Dataset archived on Zenodo.
Pushdown Is Not Free
Research on the cost of pushing LLM data-cleaning operations toward storage.
The Review Lottery
Preprint on peer-review noise.
SpecRead
Hardware specification comprehension benchmark. In progress.
Verification Debt
Open-source tracking of the backlog between AI-generated mathematical results and independent human verification. In progress.
Early American maritime history
1783–1815From Wartime Service to Canton: Reconstructing Early American Voyages to China from Digitized Newspapers, 1784–1800
With Yujia Liu. Submitted to the Journal of the Early Republic.
Yankees in the Sound: American Shipping through the Danish Straits, 1783–1815
With Yujia Liu. Submitted to International Journal of Maritime History.
The Neutral-Carrier Recovery, 1809–1811: Cotton, Dyewoods, and the Reorientation of the American Baltic Trade
With Yujia Liu. Submitted to Scandinavian Economic History Review.