New York University · ML systems · backend infrastructure

I build AI and backend systems that hold up under real constraints.

I am a New York-based computer science student at NYU focused on multi-cloud training workflows, backend platforms, observability, and research-driven engineering. I like systems that are measurable, reproducible, and production-minded.

Open resume Email me
LinkedIn GitHub Google Scholar CloudTune
New York, NY B.S. in Computer Science, Expected May 2027 Interested in U.S. internships and long-term software roles

About

What I bring to a team

My strongest lane sits at the intersection of product-minded software engineering and systems work: I build training orchestration, backend services, measurement pipelines, and tooling that makes complicated ML workflows easier to operate.

That means I spend time on infrastructure details that actually matter in practice: deployment ergonomics, cost controls, run lineage, telemetry, performance tuning, and the boring reliability work that makes research or product experiments usable by other people.

Current lanes

  • Multi-cloud AI platforms and GPU-aware orchestration
  • Backend systems in FastAPI, Spring Boot, Redis, and MySQL
  • 3D reconstruction and wireless simulation pipelines
  • Observability with Prometheus, Grafana, and deployment tooling
Python Java TypeScript FastAPI Spring Boot Docker Terraform AWS GCP Azure PyTorch CUDA Prometheus Grafana

Selected Work

Projects and systems I would want to discuss in an interview

CloudTune Founder & Developer

Multi-tenant AI platform across AWS, GCP, and Azure

Built and deployed a platform for dataset intake, training orchestration, model registry, evaluation, and endpoint operations.

  • Built the control plane with FastAPI, Docker, Terraform, CI/CD, and Prometheus/Grafana.
  • Added GPU-aware scheduling, cost accounting, and release-control workflows.
  • Validated flows from signup and run registration through evidence export and inference.
Visit CloudTune
NYU Wireless Research

Material-aware 3D reconstruction for wireless simulation

Built an end-to-end reconstruction pipeline for indoor and outdoor scenes used in computer vision and wireless modeling workflows.

  • Reconstructed about 2,500 sq m of indoor space with under 10 cm geometric error and 89% material precision.
  • Improved NYURay simulation accuracy by 18% through ML-enhanced scene and material modeling.
  • Extended workflows to 2 sq km outdoor environments with plane detection, object detection, and large-scale point-cloud processing.
ArchAI AI Research Intern

Multimodal generation workflows for internal research and demos

Built and optimized ComfyUI-based workflows for multimodal generation and text-to-video experimentation.

  • Developed internal pipelines spanning diffusion, retrieval, and LLM-assisted content generation.
  • Benchmarked latency, cost, and usability trade-offs across open-weight model families.
  • Focused on practical workflow design rather than isolated model demos.
Nanjing HuiJin Tech Backend Systems Intern

AI-assisted backend monitoring and performance tuning

Worked on Java and Spring Boot backend systems with measurable reliability and performance improvements.

  • Reduced system downtime by 15% with AI-assisted monitoring services.
  • Increased throughput by 40% through multithreading and Redis caching.
  • Reduced API latency by 25% with MySQL query optimization while maintaining 99.9% availability.

Also built

Axiom for AI evidence and audit rails, a realtime trading engine for market data experimentation, and several student/startup projects that taught me how to ship quickly without ignoring systems quality.

Experience

Roles with real engineering ownership

CloudTune · Founder & Developer

Jan 2025 - Present · New York, NY

Leading product and infrastructure work for an AI platform that spans training, evaluation, release workflows, and deployment operations across three clouds.

ArchAI · AI Research Intern

May 2025 - Aug 2025 · Beijing, China

Built multimodal generation workflows, benchmarked open-weight model families, and improved the internal reliability of research/demo pipelines.

NYU Wireless · Teaching Assistant, ECE-1002

Feb 2025 - May 2025 · New York, NY

Supported 160+ students through labs and office hours, designed ABET-aligned assignments, and contributed instructional material for the course.

Nanjing HuiJin Tech Company · AI-Driven Backend Systems Intern

Jan 2023 - Dec 2023 · Nanjing, China

Shipped backend monitoring services and performance improvements in Java, Spring Boot, Redis, and MySQL with measurable impact on throughput, latency, and uptime.

Research and Proof

Research-backed engineering, not research for its own sake

Research highlight

NYU Wireless + computer vision

My research work is most useful when it turns into infrastructure: cleaner datasets, better reconstruction pipelines, and more accurate simulations that other people can reuse.

  • FR3 measurement campaigns across 100+ TX-RX pairs with structured data capture.
  • Wireless simulation support through scene and material modeling.
  • Practical tooling around COLMAP, Open3D, NYUSIM, and large-scale geometry processing.

Selected publications

IEEE ICC 2025

  • Urban Outdoor Propagation Measurements and Channel Models at 6.75 GHz and 16.95 GHz for 5G and 6G.
  • Upper Mid-Band Channel Measurements and Characterization at 6.75 GHz and 16.95 GHz in an Indoor Factory Scenario.
View full publication list

Leadership and certifications

Evidence that I ship outside class projects

  • Hosted 6 hackathons with 200+ total participants around AI and engineering.
  • 1st Place at Pulse NYC Hackathon 2025 for an LLM-backed product prototype.
  • AWS Solutions Architect Professional, AWS Security Specialty, AWS Solutions Architect Associate, and Google Professional Data Engineer.

Notes

Build logs and technical writing

CloudTune Build log

Distributed GPU broker playbook

Notes on Terraform modules, market-aware scheduling, checkpointing, and observability for multi-cloud training infrastructure.

Read note
Axiom Systems essay

Compliance rails for fast AI teams

How evidence capture, binder exports, and traceable workflows fit underneath AI platforms and regulated engineering environments.

Read note
Trading engine Architecture note

Realtime systems and execution controls

Architecture notes on market data ingestion, hybrid signals, observability, and safe execution for experimentation in realtime systems.

Read note

Contact

Reach out if you want someone who likes hard systems problems

I am most interested in ML systems, backend infrastructure, cloud platforms, and research-heavy engineering teams. Email is the fastest path. LinkedIn works if you are recruiting.

yw7566@nyu.edu +1 (332) 280-2091 linkedin.com/in/yanzewu github.com/David-Wu1119

This form stays static and opens your default mail client with a prefilled draft. If mailto is blocked, use yw7566@nyu.edu.