bijux-atlas
bijux-atlas is a Rust-owned genomics dataset delivery platform for validated GFF3 and FASTA ingest, immutable gene-query artifacts, governed HTTP APIs, and reproducible operations evidence across local, kind, and Kubernetes deployments.
bijux is the public repository and documentation surface for projects by Bijan Mousavi across platform engineering, control-plane design, Rust and Python runtime systems, workflow execution, bioinformatics, data infrastructure, and scientific computing.
This homepage is organized around the real public surfaces: source repositories, published Python packages, published Rust crates, project documentation, and course books.
The portfolio is strongest in software where command interfaces, release lines, deployment paths, workflow behavior, and execution evidence have to remain understandable under operational pressure.
The flagship systems are the broadest public entry points. The focused repositories are narrower by scope, not by seriousness: they each concentrate on a hard problem in retrieval, replay, vector execution, document processing, graph lineage, or GNSS receiver work.
bijux-atlas is a Rust-owned genomics dataset delivery platform for validated GFF3 and FASTA ingest, immutable gene-query artifacts, governed HTTP APIs, and reproducible operations evidence across local, kind, and Kubernetes deployments.
bijux-cli is a Rust command runtime with routed plugins, structured machine-readable output, REPL and CLI parity, runtime inspection commands, and a Python compatibility bridge for the same underlying runtime.
Agentic Flows is a workflow execution engine for offline runs where replay, invariant checks, execution budgets, and explicit run records matter. Runs are meant to be inspected, replayed, and compared instead of treated as opaque automation.
Agentic Proteins is a protein design runtime that makes seeded search, run configuration, artifact capture, and shared CLI/API contracts explicit. It is built for research workflows where stochastic exploration still needs stable inputs, preserved outputs, and inspectable runs.
bijux-dna is a private bioinformatics execution system with governed tool admission, digest-pinned containers, FASTQ to BAM and VCF stage composition, policy-gated CI, and HPC-portable operation. It is designed as a reusable execution substrate for genomics workflows rather than a single one-off pipeline.
bijux-rag is a Python retrieval toolkit with CLI, FastAPI, and library entry points for chunking, indexing, querying, and serving retrieval pipelines. The package emphasizes typed adapters, testable transformations, streaming, retries, and deployable Python packaging.
bijux-rar is a retrieval-augmented reasoning system for runs that need to be checked, replayed, and verified. Replayable traces, immutable run artifacts, and execution evidence are part of the runtime contract rather than after-the-fact observability.
bijux-vex separates exact, replayable vector execution from approximate ANN search and makes approximation visible in the run record. Provenance, replay boundaries, and query behavior are explicit parts of the execution surface.
bijux-agent is a document-processing runtime for files and text with structured outputs, run traces, a CLI, and an optional HTTP API. It packages auditable text processing, explicit inputs, and checksum-guarded reference material into a small but complete public system.
bijux-dag focuses on graph execution evidence: stable graph identity, run identity, artifact lineage, replay classification, and semantic diff between runs. It is built for the question of what changed, not just whether a run finished.
bijux-gnss is a GNSS receiver and geospatial systems repository spanning acquisition, tracking loops, observation processing, navigation, and RTK/PPP-oriented foundations.
These course books cover Python functional programming, Python metaprogramming, build systems, and workflow engineering. Examples, capstones, and repository state move with the text so the material stays close to real build, runtime, and workflow problems.
A ten-module course on functional programming in Python covering purity, effect handling, async streaming, algebraic modeling, and software design tradeoffs, grounded in a real evolving codebase.
A course on Python metaprogramming for engineers who need to understand descriptors, decorators, metaclasses, import behavior, and framework-level runtime mechanics.
A course book on GNU Make as a serious build-graph tool covering truthful dependencies, atomic outputs, parallel safety, verification, diagnostics, and a runnable capstone.
A workflow-engineering course book on explicit inputs and outputs, validated configuration, safe dynamic behavior, artifact checks, and executable reproducibility.
bijux is the public engineering portfolio and repository namespace maintained by Bijan Mousavi.