Skip to content
View edsonsims's full-sized avatar

Sponsoring

@gaborbernat
@pallets

Block or report edsonsims

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
edsonsims/README.md
Edson Henrique — Montréal

Hey, I'm Edson. 👋

I'm a software engineer in Montréal, with a background in backend systems, cloud infrastructure and performance at CapIntel, SSENSE and XP Inc. Here you'll find my open-source work on service architecture, retrieval and local decision models.

🧪 Open work

Architecture Studies — one document service in three backend designs. Node.js and Python APIs, SQLite and integration tests. Decision Workbench — local decision models, public data and visible tradeoffs, with reproducible evaluations, typed APIs and bounded workflows. Enterprise AI Lab — tenant-aware retrieval compared with an unsafe baseline, using offline fixtures and reproducible reports.

Merged upstream contributions

  • pyproject-api — Prevent source-tree deletion from unsafe metadata output paths, with regression tests for ancestor paths and symlink aliases.
  • pytest-env — Honor explicitly empty environment-file lists instead of loading fallback files, with subprocess regression tests.
  • filelock — Handle NetBSD symlink-refusal errors as lost ownership in soft reader/writer locks.
  • pyfakefs — Measure setup benchmarks with CPU time after warmup, retaining coarse regression limits and testing the measurement itself.

🎙️ The story behind UHIRED

I knew my experience. Explaining it in an interview was harder. That's why I built UHIRED: practice answering questions aloud around a job description, then get feedback on each answer.

It won 1st place, Phase 2 — Learn and Integrate, in the Microsoft Developers AI Learning Hackathon (2024).

The project & demo · Microsoft's announcement

Elsewhere

LinkedIn ↗ · Devpost ↗

The concrete should serve the abstract. The temporal to the eternal.

Pinned Loading

  1. architecture-studies architecture-studies Public

    A Node.js and Python document-inspection service in three architectures, with SQLite, HTTP APIs, integration tests and generated state sequences.

    Python

  2. decision-workbench decision-workbench Public

    Local decision models on public legal and banking data, reproducible evaluations, option-order probes, typed APIs and bounded review workflows.

    Python

  3. enterprise-ai-lab enterprise-ai-lab Public

    Approval-gated RAG: public PDFs and OCR, tenant/role permissions, source citations, Temporal crash recovery and reproducible local experiments

    Python