reproducible-research

Open-source, local-first desktop AI research workbench for scientific computing with Python/R, MCP bioinformatics tools, SSH/WSL/GPU runtimes, and OpenAI/Anthropic models.

Deterministic, resource-governed per-locus genomics analyses with transactional artifacts, receipts, replay, Arrow, and reproducible sharding.

Deterministic optical-AI front end in pure Rust: a learned phase mask compresses light into task-useful sensor pixels.

A framework for defining, building, and executing reproducible HPC experiments using Nix.

Distributed integrity infrastructure for collusion-resistant independent verification. Blind peer validation of any verifiable claim — scientific, regulatory, or institutional — recorded as tamper-evident Harmony Records. Built in Rust on Holochain.

Using functions and auctions for a sustainable Fog model

Open, validated, reproducible PNT-resilience simulator spanning the full positioning, navigation & timing stack — orbit, clocks, INS, GNSS, integrity, resilience, lunar/cislunar & deep-space nav — plus quantum-vs-classical models. Runs as a Rust lib/CLI, Python, in-browser (WASM), an MCP server / Claude Code plugin, and a JetBrains plugin.

Distributed Micro Network Emulation (DµNE) Framework

Browser-native quantum chemistry: RHF, LDA, B3LYP, B3LYP-D3(BJ) with analytical gradients, geometry optimization, PES scans, Hessians, IR/Raman spectra, and RRHO thermochemistry. Pure Rust → WebAssembly, validated against PySCF 2.11.0 across 108 systems with sub-microhartree accuracy.

Portable, local-first memory layer for AI coding agents. One .agent/ folder shared across Claude Code, Cursor and Cline over MCP. Plain files, no vector DB.

A research agent for scientific discovery — part of Aviskaar's applied AI research suite.

FROST-GKR: Frobenius Reduction Over Shifted Tables for batched Poseidon2b relations over binary tower fields.

A CPU benchmark suite that shows its work. Six workloads, calibrated measurement windows, barrier-synchronised threads, robust statistics, full machine provenance, and Ed25519-signed results.

Test local AI models like a scientist: ground-truth benchmarks for LM Studio + cloud LLMs. Vision/tools/reasoning/security, N=3 trials, SHA3-sealed evidence, live Rust+SSE dashboard. No marketing numbers — your hardware, your proof.

Reproducible R source context for coding agents

CPU-first Rust research implementation of OpenAI GPT-OSS with exact MXFP4 kernels, reproducible benchmarks, and archived accelerator research.

Weft — the execution ledger for autonomous coding agents. Verification, not human review, is the merge gate. Paper: doi.org/10.5281/zenodo.21882499

Repository for creating and running benchmarks around the static analysis task of discovering usages of certain code units.

Durable, source-bound AI policy-decision receipts with crash recovery, signed Merkle epochs, and reproducible evidence.