Mechanical Engineer — design, simulation, collaboration & complex engineering systems.
Turning complex engineering constraints into designs that are clear, reliable, and buildable, and exploring new ways of working in which an AI agent collaborates with real hardware systems.
Exploring how an AI agent works with a real 3-axis motion platform: through safety gating and deterministic execution, a workflow loop driven by natural-language interaction runs from image input to physical plot.
Taylor-Couette flow is a classic model for studying rotating shear flow, flow instability, and the evolution of vortex structures. The project built a repeatable, measurable experimental platform to support comparative flow-field studies of Newtonian fluids and shear-thinning non-Newtonian fluids.
A large estuary tidal gate must provide storm-surge and flood protection under extreme tide levels while keeping everyday navigation open, making it a typical large, composite engineering system. Under multiple constraints spanning flood control, navigation, hydrodynamics, fabrication, maintenance, and operation, the project evaluated an engineering solution that balances safety, feasibility, and maintainability.
Sanitized — no original drawings or engineering parameters
The Three Gorges ship lock is the inland-waterway ship lock with the most consecutive chambers and the highest total head and inter-chamber filling head among ship locks built to date. To improve the management and O&M efficiency of complex mechanical facilities, the project established a unified, traceable digital-asset foundation and a visualization platform.
Sanitized — no original drawings or engineering parameters
Combining high-speed imaging experiments with Fluent numerical simulation to study cavity evolution and the transient hydrodynamic response during the water entry of a hydrophobic-surface sphere.
The underbody diffuser improves the airflow and pressure distribution beneath the car to raise its downforce and aerodynamic efficiency. Combining 2D concept screening with simplified 3D full-car simulation, the project made an engineering trade-off between aerodynamic gain and manufacturing complexity.
Clarity is a scarce engineering skill. Turning vague requirements into explicit goals, constraints, and outputs matters more now that people and AI agents are starting to learn how to work together.
AI is not a wish-granting machine. Only by learning where its limits lie through steady, real use can people, AI, and traditional tools each take on the work they suit.
Let constraints pick the tool. A handy tool, used long enough, quietly starts defining the problem for you.
Collaboration ends in an actionable consensus; who was right matters less than reaching one.
Thanks for taking the time to look through my work.
If anything here feels interesting or relevant to what you're building, feel free to reach out — I respond more reliably than an underconstrained simulation.
The water entry of a hydrophobic-surface sphere triggers free-surface breakup, air entrainment, and cavity closure, making it a typical strongly transient multiphase-flow problem. Combining high-speed imaging experiments with numerical simulation, the study examines cavity evolution and the transient hydrodynamic response across different entry velocities and sphere density ratios.
Experiments recorded the sphere's vertical water entry and cavity evolution with high-speed imaging. The numerical computation captured the gas-liquid interface with VOF, coupled an SST turbulence model, and represented the hydrophobic-surface effect through CSF and a contact-angle setting. A locally refined structured mesh with dynamic layering handled the sphere motion, and a custom UDF defined parameters such as entry velocity and sphere density for the different cases.
For a representative deep-seal case, the experiment-simulation error in pinch-off time, pinch-off location, and sphere-center depth at pinch-off all stays below 5%. The numerical study further groups the cavities into three typical types: quasi-static, deep-seal, and surface-seal, and analyzes how entry velocity and sphere density ratio affect cavity evolution and the transient hydrodynamic response.
Built a 2D axisymmetric Fluent VOF model; completed the structured mesh and solver setup, case runs, and post-processing of transient results; contributed to building the high-speed imaging setup and handled the experiments and data acquisition.
Breaking a real flow problem into two research tracks, numerical computation and experimental testing, and cross-validating the model and conclusions through comparison and parameter studies.
Centered on the underbody diffuser of the Silver Shark VII car, the project weighed downforce, drag, body packaging, and manufacturing constraints together. Four 2D diffuser profiles were first compared to screen out the one with the better overall performance; the candidate profile was then placed into a simplified full-car model to evaluate the 3D flow field produced by the diffuser and body together.
Adding a rear side-channel gave only limited aerodynamic gain while raising packaging and manufacturing complexity, so the project ultimately dropped that feature. 2D screening narrowed the design space, and the computation was then invested in 3D full-car evaluation, providing the basis for the design trade-off.
Ran the 2D and 3D CFD analysis of the diffuser; handled geometry cleanup, meshing, Fluent setup, post-processing, and technical-report writing.