English | Español
Software & Cyber-Physical Systems Developer
Python & Rust · Low-level systems, distributed systems, embedded computing, cloud infrastructure
I build high-performance systems with a focus on security, software verification, and reliability in production — from firmware on a microcontroller to control planes running on Azure.
I design and ship low-level and distributed systems: embedded computing, cloud infrastructure, and language/codegen tooling, mostly in Python and Rust. My work centers on one question — does this actually hold up in production? — so every generator, transpiler, or runtime I publish is checked against real toolchains and real deployments, not just unit tests on strings.
I'm also a technical instructor, open-source mentor, and speaker at tech community events (Microsoft Build 2026 Community Event, CSWeek 2026).
Background in International Business, which helps me translate business needs into systems that ship.
Python · Azure Functions · PyPI · MCP · Paddle
A serverless runtime exposing Email, Telegram, WhatsApp, Discord, Slack, Google Sheets, HTTP, Webhooks, and CloudScheduler through a clean Python SDK, plus an AI module (agentic tool-use, structured extraction, stateless RAG) that reuses the MCP tool catalog. Published on PyPI and validated by a 61-check end-to-end suite run against a live production deployment — with per-tenant HMAC key derivation, ContextVar isolation, and process-level sandboxing.
Python (ast) · Rust (PyO3) · Java · Go · JavaScript · Rust
Full 16/16 source-to-target coverage across Java, Go, JavaScript, and Rust, using Python as a universal AST pivot. The Rust emitter is ownership-aware (exact .clone()/.to_string(), Option<T>, div_euclid/rem_euclid for Python semantics) — the borrow checker ends up being the pipeline's strictest reviewer. Every output is verified by actually compiling and running it with rustc and javac, backed by eight rustc-style lints and a dual-engine safety analyzer.
Python · Rust · embedded-hal · 5 MCU toolchains
A single typed Python API (IOConfig) describes hardware once and generates native firmware for five different toolchains — no more full HAL rewrites when porting between MCU families. Design-time validation rejects electrically invalid pin/peripheral configurations before emitting code, with rustc-style diagnostics localized into the user's language. Verified by generated Rust compiling under thumbv6m-none-eabi and 343 tests.
Rust · axum · SeaORM · PostgreSQL/SQLite
An all-Rust Cargo workspace providing accounts, identity, service catalog, and per-account activations as a control plane kept out of the request path. Architectural boundaries are enforced at the crate level — the identity logic knows nothing about axum or SeaORM. Tokens are 256-bit CSPRNG, stored only as SHA-256 hashes, with a test verifying the plaintext token is never a store key.
Python · stdlib-only · MIT
Turns opinion-based capacity planning into measurement: a fluent API models pipelines as stages with capacity and latency, locates the bottleneck, prescribes the minimal change to hit a target rate, and emits a shareable Architecture Decision Report. 77 tests, zero dependencies.
ESP32 · Raspberry Pi · Kalman Filter · MQTT · SQLite · SSE
A three-node cyber-physical system: discrete-time Kalman filtering on the MCU, an FSM-controlled multi-sensor pipeline (PIR + ultrasonic), UART → MQTT → edge distribution, SQLite persistence, a REST API, and a live dashboard over Server-Sent Events.
Repositories exploring the core engineering principles behind robotics and cyber-physical systems — sensing, estimation, control, and distributed coordination under real-world constraints.
- sensor-uncertainty-lab — Probabilistic models of noisy sensor measurements
- bayesian-sensor-fusion — Kalman filters, particle filters, and multi-sensor fusion for state estimation
- robot-perception-lab — Probabilistic perception: occupancy grids, localization
- control-systems-lab — PID controllers and system dynamics
- embedded-state-machine-systems — FSM architectures for embedded robotics
- edge-device-coordination — Coordination patterns for distributed embedded nodes
| Area | Primary use |
|---|---|
| 🐍 Python | Automation, backend APIs, language engineering (AST, codegen), data pipelines |
| 🦀 Rust | Systems, embedded, embedded-hal, safe low-level tooling, web services (axum) |
| ⚙️ C/C++ | Arduino, ESP-IDF, STM32 (bare-metal / HAL), proprietary SDK integration |
| 🔌 Embedded & IoT | ESP32, STM32, Arduino, RP2040; PlatformIO, ESP-IDF, STM32Cube; UART, I2C, SPI, MQTT |
| ☁️ Cloud | Azure Functions (production), core AWS services (EC2, Lambda, S3, IAM), serverless multi-tenant design |
| 🗄️ Data & storage | PostgreSQL, SQLite, SeaORM, SQLAlchemy |
| 🤖 AI Agent Systems | MCP server design, agentic tool-use, JSON-RPC 2.0 |
- "Building a Multi-Tenant Python Runtime on Azure Functions" — Microsoft Build 2026 Community Event, Azure User Group Latam (Lima, June 2026)
- "Isolation and Trust Boundaries in Production: Building Failure-Proof Systems" — CSWeek 2026 (Lima, August 2026)
- Author of The Agnostic Engineer: Architecture Beyond Infrastructure (complete manuscript, unpublished) — on designing systems that survive changes in infrastructure, language, and provider
- 📧 Email: frostcore@jafa.dev
- 🌐 jafa.dev
- 💡 Open to collaboration on embedded systems, industrial automation, and hardware–software integration
- 🌎 Available for remote consulting and technical advisory roles with international teams
