Skip to content
View VictorMartins3's full-sized avatar
4 followers · 7 following

Achievements

Achievement: Pull Sharkx2

Achievements

Achievement: Pull Sharkx2

Highlights

  • Pro

Block or report VictorMartins3

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
VictorMartins3/README.md

Victor Martins

Software engineer · Distributed systems · Security · Performance

LinkedIn · X / Twitter · Email

I'm a software engineer based in São Paulo, Brazil. My background is in distributed backend systems, payment infrastructure and security engineering, working primarily with Rust and TypeScript.

I've worked on payment services, identity verification, authentication and event-driven processing, from implementation through deployment and production debugging. I also have a background in robotics and mechatronics.

My open-source work focuses on dependency analysis and reproducible benchmarks for systems performance.

Engineering background

  • Backend & infrastructure: distributed services, APIs and asynchronous workflows; PostgreSQL, Redis, AWS, GCP, Kubernetes and Terraform.
  • Security: authentication, cryptographic workflows and identity verification, with experience in payment and fraud-prevention systems.
  • Performance & reliability: profiling, concurrency, observability and debugging across application, network and operating-system layers.

Working on

capsurface an offline tool for reviewing npm dependency updates.

It compares source scans against a reviewed baseline and points to newly observed filesystem, network, process-execution and credential-access indicators. The goal is to give reviewers a useful before/after, with source locations they can inspect locally or in a pull request.

It's early. I'm especially interested in real updates where the report is noisy, misses a change, or leaves you unsure what to review next.

Try the demo · Report an issue

Performance experiments

I also write benchmarks to understand where time goes, from memory access to compiler behavior.

Project What it explores
roofline-bench Memory bandwidth and arithmetic intensity in CPU workloads modeled on LLM inference. C.
cache-bench How matrix traversal order and data layout affect CPU cache behavior. C.
mc-greeks-bench Finite differences, hand-written differentiation and compiler-generated derivatives for Monte Carlo option Greeks. Rust.
trait-tax Dynamic dispatch versus a static processing pipeline. Rust.

Get in touch

For bugs, examples or technical questions, open an issue in the relevant repository. You can also reach me at martins.victor212004@gmail.com.

Popular repositories Loading

  1. capsurface capsurface Public

    A capability-aware supply-chain scanner for npm. Diffs each dependency's capability surface against a reviewed baseline and fails CI on escalation. Zero dependencies, offline.

    JavaScript 6 1

  2. cache-bench cache-bench Public

    A hands-on lesson in CPU cache behavior. Allocate a matrix, traverse it two ways, see why one flies and one crawls. Prove to yourself that data layout beats SIMD

    C 2

  3. payment-engine payment-engine Public

    payment engine test case

    Rust

  4. Enzyme Enzyme Public

    Forked from EnzymeAD/Enzyme

    High-performance automatic differentiation of LLVM and MLIR.

    LLVM

  5. mc-greeks-bench mc-greeks-bench Public

    Benchmark comparing finite differences, hand-written AAD and Rust std::autodiff (Enzyme) for Monte Carlo option Greeks

    Rust

  6. roofline-bench roofline-bench Public

    Benchmark: why LLM token generation is bound by memory bandwidth, not compute. Batch-size sweep over one weight matrix, with the bandwidth ceiling measured on the same machine.

    C