skyler@uiuc:~$ whoami
Computer Engineer — Illinois, class of 2027
I build the layer where software meets hardware.
> currently_compiling:
This past year I built digital twins of AI compute clusters at Lenovo, shipped computer vision to a live factory floor in Taipei — and now I'm launching Hoop AI, my own computer-vision shot tracker. Consider this my digital handshake — firm, no sweaty palms.
Open to full-time SWE / systems roles — starting 2027About
// the short version
I'm a Computer Engineering student at the University of Illinois Urbana-Champaign who likes working at the boundaries — where a Python script has to convince a switch to route packets, where a neural net has to fit on an edge device bolted to a factory line, where an analog circuit has to sound good through a guitar amp.
My favorite kind of problem is the one nobody has scoped yet. At Lenovo I inherited a half-finished internal tool from an engineer who had left the company, and turned it into a working pipeline that spins up virtual copies of entire AI datacenter networks. At JPC in Taipei, I pitched two AI systems that didn't exist before I proposed them — then built both. Now I'm doing it for myself: Hoop AI, my computer-vision shot tracker, is live for pre-launch reservations.
I care about proof over polish: LLDP tables, converged BGP sessions, TTL-verified pings, oscilloscope traces. If I claim something works, I can show you the evidence. That extends to how I work — I use AI tools aggressively in my workflow, and verify everything they produce, because I've watched them be confidently wrong.
B.S. Computer Engineering — University of Illinois Urbana-Champaign, expected 2027
Coursework: Computer Systems & Programming, Digital Systems Lab, Data Structures & Algorithms, Signal Processing, Robotics
Network Automation
Digital twins of datacenter fabrics, BGP/FRR config generation, topology verification, and infrastructure-as-code.
Embedded Systems
PCB design through firmware: microcontrollers, FPGAs, motor control, sensor integration, and debugging with real test equipment.
Applied ML
ML that ships: RAG systems on private data, real-time vision on edge devices, and audio DSP models — deployed, not just demoed.
Experience
// git log --oneline --reverse=false
Lenovo — Network Automation Engineer Intern
- Sole owner of an internal Python pipeline that converts AI compute-cluster network designs into deployable containerlab digital twins — virtual SONiC switch fabrics that let engineers validate a cluster's entire control plane before a single piece of hardware is purchased.
- Automated the full routing layer: numbered eBGP over FRR with structured 4-byte ASN assignment, surgical config_db patching on SONiC, and automatic server addressing — zero manual configuration from design file to converged fabric.
- Proved end-to-end correctness on all four routable fabrics (GPU backend, frontend, and two storage networks): LLDP topology verification, BGP session establishment, and real server-to-server traffic confirmed by TTL-decrement analysis across leaf→core→leaf paths.
- Inherited legacy topology code from a departed engineer with full design authority; made and documented the core fidelity decision (structural fidelity over proportional scale-down, after proving exact proportional shrink mathematically impossible for real designs).
JPC — Jess-Link Products Connectivity — AI Engineer Intern
- Proposed and built a RAG document-search system (LangChain + ChromaDB + Ollama, fully on-prem) so engineers could query proprietary engineering documents in natural language — cut average search time by over 50%.
- Initiated an AI defect-detection pipeline using YOLO and Mask R-CNN, optimized for the NVIDIA Jetson Orin NX and integrated with factory systems for real-time quality control on the production line.
- Both projects were self-pitched: identified the problem, made the case to management, and delivered working systems within one summer.
iRobotics (Micromouse) — Electrical Team Lead
- Lead the electrical team building an autonomous maze-solving robot: schematic and PCB design, sensor interfacing, power management, and PID-controlled motors.
- Contributed a modified flood-fill pathfinding algorithm that maps unknown mazes and computes the fastest escape path under real hardware constraints.
eFold — Embedded Systems Engineer
- Student-led team building a modular electric skateboard from scratch — motor control (VESC), battery management, and regenerative braking.
- Modified VESC hardware and firmware to support a 12S2P battery pack, improving battery life by 17% via MOSFET and voltage-divider redesigns for safe high-voltage operation.
Illini Solar Car — Electrical Engineer
- Built and tested circuits for Calypso, a road-going solar-powered race car; designed an improved braking sensor using a custom string potentiometer for finer control accuracy.
Byte-Sized Coders — Cofounder & VP
2018 — 2023 · TaipeiFounded a coding-education club with 30+ active students. Wrote and taught a year-long Python curriculum, and ran Scratch summer camps at underprivileged elementary schools.
FIRST Robotics — Software Developer
2020 — 2022 · TaipeiWrote autonomous-mode code (navigate, shoot, climb — no human input) and integrated sensor feedback with motor control for competition robots.
Projects
// things that actually run
AI Cluster Digital Twin Pipeline
A Python pipeline that takes a paper design of an AI datacenter network — hundreds of switches, five separate fabrics — and boots a faithful, scaled-down virtual replica of it in containerlab: real SONiC network operating systems, real BGP converging over FRR, real packets crossing a leaf-spine fabric. Design flaws surface in a lab VM instead of on a customer's loading dock.
// internal tool — code private, whiteboard walkthrough gladly offered in interviews
Hoop AI — Every make. Every miss. Counted.
Prop a phone on the sideline, shoot your session, and on-device computer vision counts every shot — no wearables, no sensors, no hardware. Trained where people actually play: outdoor asphalt, chain nets, no nets, bad light. I own the whole thing — YOLOv8 vision model, training data, the trilingual landing site, and the Supabase/Vercel signup backend.
// built solo: CV model + mobile pipeline + landing page + payments-free waitlist
Engineering-Docs RAG Search
On-prem retrieval-augmented search over proprietary engineering documents. Natural-language questions, grounded answers, zero data leaving the building.
Real-Time Defect Detection
Vision pipeline spotting product defects on a live production line — YOLO for detection, Mask R-CNN for segmentation, squeezed onto a Jetson Orin NX at the edge.
AI Guitar Effects Pedal
A pedal that listens to a reference guitar tone and replicates it — MFCC + OpenL3 embeddings feed a CNN that identifies the effect and estimates the knob settings, live on a Raspberry Pi.
FPGA Image Pipeline
Reads raw images off a micro-SD card over SPI, stages pixels through BRAM, and drives an HDMI display with a custom VGA controller — a complete image pipeline with no CPU anywhere.
Wireless Guitar System
RF transmitter/receiver pair replacing a guitar cable: ~10 ms ADC latency (MCP3008), low-noise NE5534 amplification, custom PCB — prototyped on breadboard, verified on oscilloscope, laid out in KiCad.
Analog Distortion Pedal
Fully analog distortion from first principles: op-amp gain staging, selectable clipping diodes (LED / germanium / Schottky — each clips differently), and a tuned tone stack.
Beyond the terminal
// proof I'm a person, not a resume-generating script
I've finished a half-Ironman
113 kilometers: 1.9 km swim, 90 km bike, 21.1 km run. Months of training for one race day taught me the skill I use most in engineering — showing up every day for a goal that's still months away. Race day is just the deploy.
70.3 finisher · 1.9 swim / 90 bike / 21.1 run (km)I teach — code, and jump shots
I cofounded a coding club in Taipei and taught a year-long Python curriculum to 30+ students, ran Scratch camps at underprivileged elementary schools, and published a basketball course that 1,500+ strangers on the internet took and rated 4.4★. If I can explain a closeout to a 12-year-old, I can explain a design review to anyone.
1,500+ students on Udemy · 4.4★I build the gear I play through
Three of the projects above exist because I play guitar and got curious about what was inside my pedals. Now my signal chain is increasingly things I soldered myself — which means when it hums, it's personal.
analog + digital + ML, all in one signal chainTaipei ⇄ Urbana-Champaign
I grew up in Taipei and study in Illinois, so I default to building for more than one audience — Hoop AI launches in English, 繁體中文, and 简体中文 on day one. Comfortable being the bridge between teams, languages, and time zones.
bilingual · products ship in three scripts05. // what's next
Let's build something real.
I graduate in 2027 and I'm looking for full-time software and systems engineering roles. If you're hiring people who are comfortable everywhere from BGP configs to breadboards, my inbox is open — I answer fast.