Spain/EU ⇄ US/MX/Americas
César Antonio Pérez Quintana · Computational physics & HPC · Backend & distributed systems
Senior backend engineer and computational physicist with ten years of experience spanning numerical ocean modelling on HPC clusters and high-throughput distributed systems. Focused on performance, reliability, and reproducible computing.
Balanced flow comes from a streamfunction, so ∇·u = 0 — parcels swirl and never
pile up. Internal gravity waves add a divergent part at higher frequency.
Telling those two apart in a ~2 TB global ocean simulation was the subject of my doctoral thesis.
Pointer injects a vortex · click reverses its circulation.
01 / The bridge
Scientific computing and distributed backend systems share fundamental requirements: numerical correctness, throughput under load, and reproducible execution. I have spent ten years working across both domains, applying the same rigor whether optimising Fortran and OpenMP kernels on an HPC cluster or operating Go microservices under high traffic.
01
When re-engineering IH-TESEO's particle core, I implemented automated regression tests for weathering processes and integrated them into GitHub Actions, ensuring numerical integrity is verified on every commit.
02
Whether handling thousands of ticket purchases per minute, hundreds of asset validations per second, tens of gigabytes of ocean data daily, or ~20k flood scenarios on a cluster, reliable throughput comes from identifying bottlenecks and profiling early.
03
Containerising numerical models like ROMS, Delft3D, and SFINCS ensures simulations are portable and results remain defensible over time. Consistent environments make deployments predictable and scientific results reliable.
04
Experience taking systems from architecture and development through Docker deployment, CI/CD, and operational maintenance—including unattended data pipelines, alerting, and internal support tools.
02 / Results, measured
Selected figures from systems I have designed, built, and operated across high-performance computing, operational data engineering, and backend infrastructure.
0×
more particles simulated
IH-TESEO pollutant transport, re-engineered Fortran/OpenMP core — with weathering physics fully validated and no loss of accuracy.
0× faster
execution speedup
Same model, up to 4× faster, plus netCDF I/O so output interoperates with standard ocean-data tooling.
0GB/day
ingested unattended
Operational forecast system producing daily ocean-current and sea-level predictions for the Cantabrian coast and Santander Bay.
~20k
HPC scenarios executed
SFINCS flood-hazard ensemble prepared and run on a SLURM cluster, with GeoPandas driving scenario generation.
~2TB
simulation output processed
Doctoral ETL pipeline over the LLC4320 high-resolution global ocean run, feeding a spectral wave–balanced filter.
O(1k)/min
peak users handled
Live-event ticketing infrastructure supported thousands of purchases per minute while validating hundreds of asset ownerships per second.
~50k
accounts onboarded
Verifiable credentials issued and validated across thousands of devices via Go microservices (REST + gRPC).
<0s
eligibility check latency
Configurable rules engine running cryptographic, logical and on-chain checks fast enough for near-real-time gate statistics.
2k
users via public SDK
Glyph's open-source SDK adopted by third-party integrations alongside internal applications.
0%
less unplanned downtime
Estimated improvement from an LSTM model predicting mass-spectrometer failure inside a scheduled maintenance window.
03 / Outstanding work
Case studies outlining problem constraints, architecture decisions, and operational outcomes.
Case 01 · IHCantabria
Fortran/OpenMP re-engineering, netCDF I/O, validated weathering physics and a CI suite guarding the numerics.
Case 05 · tokenproof
Go microservices issuing and verifying signed credentials, a configurable eligibility rules engine, and on-device ECDSA for Android and iOS.
Case 04 · Yuga Labs
Owned end to end and open-sourced: a widget plus an SDK that let third-party developers integrate multi-chain EVM accounts into their applications.
04 / The science
Interactive browser implementations of research work from my doctoral, master's, and engineering theses — demonstrating wave–vortex decomposition, Penning trap dynamics, and spectral clustering running in real time.
Ph.D. · CICESE
2 TB of a global ocean simulation, filtered in wavelength–frequency space to split gravity waves from geostrophic flow. Drag the latitude slider and watch the partition move with the Coriolis frequency.
M.Sc. · CINVESTAV-IPN
Coherent states, ladder operators and eigenvalues, derived analytically and checked numerically. The demo integrates the Lorentz equations live so you can destabilise the trap yourself.
B.Eng. · UPIITA-IPN
A quantum variant of spectral clustering with constant theoretical runtime. The demo builds the graph Laplacian, extracts the Fiedler vector, and shows k-means losing to it on the same points.
05 / Availability
This is most useful where a hard problem needs owning end to end — architecture, implementation, deployment and the day-2 reality — and especially where scientific or numerical correctness is part of the product. Based in Santander, working remotely, with hours that overlap both Europe and the Americas.