SYSTEMS / PROGRAMMING

Anthony Cristian

Systems & Data Infrastructure

View Work

DATA / INFRASTRUCTURE

Scientific Computing

Brazil — OPEN TO WORK/COLLAB

Scroll

01 — PROFILE

Background

Computer Engineering undergraduate at UTFPR, Brazil, exploring systems programming, data infrastructure and scientific computing through personal and academic projects.

▹Computer Engineering — UTFPR▹Brazil

02 — SELECTED WORK

Projects

01 — SYSTEMS / SCIENTIFIC COMPUTING

HelixSig

Concurrency with explicit boundaries.

A Linux-first signal-processing pipeline that makes ownership, overload and failure behavior explicit across Rust and C++.

RustC++20CXXNumPyParquet
Waveforms→Bounded workers→CXX / C++ kernel→Parquet
  • ▹Bounded queues & backpressure
  • ▹Explicit overload & shutdown semantics
  • ▹Independent NumPy validation · ASan / UBSan
Technical detail

Rust owns the buffers, queues and workers. The allocation-free C++20 numerical kernel borrows slices synchronously through CXX and retains no Rust pointers. Accepted, dropped, processed and written events are accounted for separately; failure tests exercise interrupted runs, writer errors and contained worker panics.

Academic experimental system. Benchmarks characterize the tested environment; no hard-real-time or production guarantees.

SOURCE CODE ↗EVIDENCE ↗
02 — DATA INFRASTRUCTURE

M2C Binary Engine

Legacy bytes. Typed data.

A Rust engine that compiles a supported COBOL copybook subset and converts fixed-record binary files into typed Arrow and Parquet data.

RustApache ArrowParquetCOBOL copybooks
Binary + copybook→Compiled layout→Arrow batches→Parquet parts
  • ▹EBCDIC, binary & packed-decimal decoding
  • ▹Recoverable parts, manifests & receipts
  • ▹Independent Spark / Cobrix comparison
Technical detail

The recoverable conversion path validates committed parts before resuming after process interruption. An optional, separate artifact-protection feature explores ML-KEM-768, HKDF-SHA-256 and AES-256-GCM. The project site presents architecture, reproducible local benchmarks and a verified fixture-based demonstration.

Educational prototype. Narrow copybook subset; local process-failure recovery. The site reveals checked-in output, not live Rust execution.

SOURCE CODE ↗PROJECT SITE ↗EVIDENCE ↗
03 — QUANTITATIVE EXPERIMENTATION

Quant AI Challenge

A Python research harness for data validation, temporal alignment and reproducible experiments. Hypotheses are tested with holdout discipline and a falsification-first process.

Research methodology, including negative results. No validated trading thesis or claimed investment performance.

PythonpandasNumPy
SOURCE CODE ↗CANONICAL STATUS ↗

03 — TECHNICAL FOUNDATION

Skills

LANGUAGES

  • Rust
  • C++20
  • Python
  • C
  • SQL

SYSTEMS

  • Linux / Unix
  • Concurrency
  • Bounded queues
  • Backpressure
  • Memory ownership
  • CXX FFI
  • Failure handling

DATA

  • Apache Arrow
  • Parquet
  • PostgreSQL / SQLite
  • NumPy
  • pandas
  • SciPy
  • Binary formats

ENGINEERING

  • Cargo / CMake
  • Git / GitHub Actions
  • Docker
  • Unit & property testing
  • Differential validation
  • ASan / UBSan
  • Reproducible benchmarks