Engineering Portfolio

Case Studies

Six engineering case studies that show how I work: constraints, tradeoffs, validation, and measurable outcomes. Each one is a story about solving real problems under real constraints.

MerLinPhotonic QMLMMD Loss

MerLin Photonic Generative Modeling

Single-circuit photonic generator trained with closed-form MMD² loss on the two-ring circles dataset. Key finding: good MMD² doesn't mean clean ring structure.

Held-out MMD²

0.0125

ring_mass

0.691

Params

220

JAXIQP CircuitsAnti-Concentration

IQP-MMD: Generative Quantum Machine Learning

Modular research toolkit for training IQP circuits with MMD loss. Key finding: low loss does not imply learned distribution shape.

Families

4

Datasets

6

Finding

MMD ≠ Shape

QiskitGradientsBenchmarking

Quantum Algorithm & Simulation Engineering

Production-oriented benchmark suite covering simulator benchmarking, noise scaling, gradient variance behavior, and transpilation effects.

Artifacts

4

Sweep

5×5×20

Finding

Sign flip

SimulationTestingBenchmarks

Quantum Simulator Validation

Correctness + benchmarking harness for a pure-Python statevector simulator with 167+ tests and reproducible scaling analysis.

Tests

167+

Scaling

O(2^n)

Ops

10-20 µs

HPCMPIOpenMPAVX2 SIMD

Distributed Meet-in-the-Middle Attack

Hybrid distributed system solving exponentially-scaling collision search on 512 MPI ranks across a 16-node HPC cluster.

Ranks

512

Speedup

171.6×

Size

n=40

MLNLPData Pipeline

Music & Machine Learning Pipeline

End-to-end ML pipeline studying content-only music popularity prediction across 1,550 tracks with multilingual lyric sentiment analysis.

Tracks

1,550

Best R²

0.20

Dims

384→213

Each study includes methodology, results, and reproducible artifacts.