FrameFlux · Telemetry · Portfolio · LinkedIn

I build complete web products with a strong focus on the engineering behind the interface: backend APIs, realtime systems, asynchronous processing, data, security and verification.

The two systems below are the center of this profile because they show those concerns in different ways.

Backend repository · Frontend repository

A FastAPI media-processing system built around PostgreSQL, Redis, ARQ background jobs and FFmpeg, with a Next.js frontend.

Interesting engineering surface

• Long-running media work is modeled as background jobs rather than ordinary synchronous CRUD.
• Resumable uploads expose initialization, chunk transfer, pause/resume, retry, cancellation and finalization flows.
• Upload boundaries include extension, MIME/category, size and binary-signature validation.
• Media processing covers probing, conversion, editing, splitting, clip operations, overlays, transformations and freeze-frame workflows.
• PostgreSQL stores durable application metadata while Redis/ARQ coordinates asynchronous work and progress.

flowchart LR
    Client[Browser / UI] --> API[FastAPI API]
    API --> DB[(PostgreSQL)]
    API --> Queue[Redis / ARQ]
    Queue --> Worker[Background Worker]
    Worker --> FFmpeg[FFmpeg]
    Worker --> Storage[Output / Storage]

Backend repository · Frontend repository

A FastAPI observability system that generates synthetic telemetry, detects anomalies, manages alert state, persists history in PostgreSQL and streams live events over authenticated WebSockets.

Interesting engineering surface

• Correlated synthetic signals cover CPU, memory, temperature, network throughput, requests/sec, latency and error rate.
• Rolling z-score anomaly detection produces alert severity and lifecycle state.
• Telemetry generation and PostgreSQL persistence are separated by bounded asynchronous processing.
• Authentication uses JWTs and Argon2 password hashing, with backend-enforced viewer/operator/admin authorization.
• Simulation controls make realtime state and fault-injection behavior observable from the dashboard.
• The frontend contains browser-level E2E coverage for authentication, live streaming, alerts, analytics, hosts, administration, authorization, settings, logout protection and responsive navigation.

flowchart LR
    Source[Synthetic telemetry] --> Manager[Telemetry manager]
    Manager --> Detect[Anomaly detection]
    Manager --> Persist[Async persistence]
    Persist --> DB[(PostgreSQL)]
    Detect --> Alerts[Alert lifecycle]
    Manager --> WS[Authenticated WebSocket]
    Alerts --> WS
    WS --> Dashboard[Realtime dashboard]

Heavy processing belongs behind a job boundary. FrameFlux uses ARQ workers for FFmpeg work; Telemetry isolates database persistence from the generation path.

The frontend can expose role-aware controls, but protected operations remain backend responsibilities. Telemetry's authorization model is enforced by the API.

Inputs should be rejected before they reach expensive or security-sensitive processing. FrameFlux's upload pipeline reflects that principle directly.

Unit and integration tests cover backend behavior; the Telemetry frontend also exercises complete browser journeys against the application boundary.

Backend
Python · FastAPI · PostgreSQL · SQLAlchemy · Alembic · Pydantic

Systems
Redis · ARQ · WebSockets · FFmpeg · Uvicorn

Frontend
React · Next.js · TypeScript · Vite · Tailwind CSS

Engineering
Pytest · Playwright · Docker · GitHub Actions

Realtime delivery · asynchronous processing · backend API design · authentication/authorization · testing confidence

For a deeper, fact-checked index of the profile's technical claims:

Open the engineering evidence map →

_{Full-Stack Engineer · Backend & Systems}