Jacob Fiske | ML, Automation

I am a forever learner building practical software across machine learning, automation, and backend systems. I focus on problem solving, debugging, and reliable implementation by designing, testing, and iterating on end to end workflows.

github: @Jacobjfiske

InferFlow

open source
InferFlow architecture visual

Queue based ML inference service.

An asynchronous ML inference service for text classification workloads. InferFlow separates request intake from execution through background job queues and durable job state, with explicit lifecycle tracking. It supports retry safe processing and status polling for reliable high concurrency workflows, and serves as the execution foundation for automation pipeline and MLOps projects.

python fastapi celery redis postgres sqlalchemy pytest github actions docker compose
View Source on GitHub

FlowLedger

open source
FlowLedger pipeline visual

Run keyed automation pipeline with reliability controls.

A scheduled ingest, transform, validate, and publish pipeline with persistent run tracking and replay safe behavior. Failed runs can be retried with the same run key, deterministic ingest errors are classified as not retryable, and CLI exit codes correctly signal failure to schedulers and CI. The project focuses on operational correctness under recurring batch workloads.

python postgres sqlalchemy pytest github actions docker compose automation
View Source on GitHub

ModelGate

open source
ModelGate lifecycle visual

MLOps template with safe model promotion controls.

A production oriented ML template that separates training and inference while enforcing explicit promotion semantics. New models land in canary by default, stable promotion is gated, runtime loading validates artifact contracts, and fallback behavior is controlled by config. CI checks and lifecycle tests target safe model rollout and rollback paths.

python numpy fastapi pydantic pytest github actions docker mlops
View Source on GitHub

Earlier Projects

SMS-Spam-Filtering

open source
SMS spam detection project media

SMS text classification service for spam detection tasks.

A supervised SMS spam classifier for labeling short text messages as spam or not spam. The implementation covers dataset preparation, feature extraction, model training, and evaluation, with focus on stable classification behavior and metric driven validation.

python scikit-learn numpy pandas ml
View Source on GitHub

Machine_Learning_Project

open source
Machine learning project media

General ML implementation for model training and evaluation.

A machine learning project covering data preparation, model training, model comparison, and performance reporting across a full workflow. It emphasizes reproducible experiments, consistent evaluation criteria, and clear presentation of model outcomes.

python scikit-learn pandas numpy jupyter ml
View Source on GitHub

Creating-And-Combining-Views

open source
SQL views project placeholder media

SwiftUI MVVM app based on Apple Landmarks patterns.

An exploration of SwiftUI using Apple Landmarks as reference, implemented with MVVM architecture and component relationships across views, lists, and navigation flows. The project includes user input handling, CoreGraphics based layout work, and integration points with UIKit.

swift swiftui uikit coregraphics ios mvvm
View Source on GitHub