About the role#
Frontier models now possess the ability to reason, write code, and use tools. The next challenge is building agents that operate reliably over long tasks using real data and deploying them efficiently. Business intelligence at TSMC scale presents a difficult environment for this, as agents must navigate messy schemas and ambiguous questions while maintaining consistent logic. We are looking for PhD and Master's students to own a research and engineering project end-to-end. You will design methods, build agent harnesses, train or adapt models, and measure progress with rigorous evaluations. The goal is to produce a first-author paper or a prototype that integrates into real decision-making workflows.
What you'll do#
- Lead a research and engineering project by framing open problems, forming hypotheses, and running experiments from small-scale ablations to full GPU training.
- Build and manage advanced machine learning, deep learning, LLM, foundation model, and time-series systems.
- Develop data-driven solutions and implement agent harnesses, tool interfaces, and environments grounded in realistic business intelligence workloads.
- Package, serve, and monitor models and agents in production-like environments while optimizing for reliability, latency, and cost.
- Analyze complex data sets and translate findings into actionable insights for business stakeholders.
- Design benchmarks and evaluation protocols to measure reliability and report negative results honestly.
- Work with international teams in the US and Taiwan to prioritize data opportunities.
- Partner with research scientists and data engineers to develop predictive models and agentic systems.
- Participate in weekly research syncs, paper reading groups, and brainstorming sessions.
- Present insights through internal reports and write up findings for publication or internal deployment.
What you'll need#
- Currently pursuing a Ph.D. or M.S. in Computer Science, Electrical Engineering, Information Systems, Information Science, Statistics, or a related AI/ML field.
- Depth in one or more areas: machine learning, deep learning, large language models, AI agents, time-series modeling, optimization, network science, or ML systems.
- Strong Python skills and hands-on experience building and shipping ML or LLM systems.
- Ability to read recent papers critically and implement or extend them quickly.
- Experience designing agent harnesses or building agents with modern SDKs and tool protocols.
- Experience with pretrained time-series foundation models, deep forecasting, or probabilistic evaluation.
- Experience with LLM serving engines, model quantization, or inference optimization.
- Experience with MLOps, deployment infrastructure, and GPU cluster environments.
- Hands-on LLM experience beyond prompting, such as fine-tuning, preference optimization, distillation, or RL post-training.
- Experience deploying models into user-facing tools like dashboarding platforms or Python and JavaScript frameworks.
- Experience with text-to-SQL, semantic parsing, tabular reasoning, or analytics.
- Contributions to open-source AI/ML projects or public benchmarks.
- Publications in leading AI conferences or strong pre-prints in relevant fields.
Location & details#
- Location: San Jose, California
- Term: Summer 2027
- Modality: On-site
- Status: Full-time, paid internship
About TSMC
TSMC operates as a dedicated semiconductor foundry. It produces wafers for integrated circuit suppliers across the globe. The company maintains multiple wafer fabs across Taiwan, the United States, China, and the Netherlands. It was established in 1987 and currently employs over 26,000 people. TSMC is a public company listed on both the Taiwan Stock Exchange and the New York Stock Exchange.
How to get in at TSMC
Applying to TSMC early provides a distinct advantage because recruiters review applications as they arrive. Intern Insider sends an instant alert the moment a role matching your target is published anywhere, so you apply among the first before the pile grows. This helps you avoid the frustration of submitting your resume into a crowded queue. You can also improve your chances by reaching out to the right people. Intern Insider surfaces the recruiters behind the company's roles so you can reach out directly to ask about the role or a referral. Connecting with a human contact often improves response rates compared to standard submissions.
