What we've learned shipping Meridian — an MCP server with real auth, Stripe billing, RAG via Vectorize, and SSE streaming on Cloudflare's free tier. Plus side notes on photonic / quantum retrieval research.
A technical briefing on arXiv:2606.20402: showing how entangling tensor kernels are Matrix Product Operator factorizations of Fourier coefficient tensors, revealing that spectral compressibility governs classical representability.
A technical briefing on arXiv:2606.19133: how SO(3)-equivariant graph neural networks capture geometric expressiveness to outperform rotation-invariant models in materials screening.
A briefing on arXiv:2606.18169: utilizing Holevo's framework and Toeplitz matrix optimization to establish absolute Heisenberg limits in quantum phase estimation.
A technical briefing on arXiv:2606.16938: Achieving simultaneous all-optical edge detection and classification using integrated spin-multiplexed nonlocal metasurfaces.
A technical briefing on arXiv:2606.14698: Confirming the globally information-optimal measurement for symmetric state ensembles through the resolution of the quantum pyramids conjecture.
A briefing on arXiv:2606.13641: introducing a unified Hamiltonian framework that leverages pairwise interactions to maximize information density in near-term quantum processors.
A briefing on arXiv:2606.12394: establishing a parallel protocol that achieves asymptotically optimal O(1/n) noise suppression for quantum gate purification.
A briefing on arXiv:2606.10253: proving that spatial multiplexing in ONNs can serve as a trainable representational coordinate, enabling million-parameter hybrid vision-language systems.
A briefing on arXiv:2606.09554: Establishing absolute physical limits for optical trapping performance. We dissect a new framework that casts trap stiffness maximization as a convex optimization problem.
A briefing on arXiv:2606.07265: computing true multi-input multi-output convolutions in a single shot using spatial micro-organization, microlens arrays, and optical sum-reduction.
A briefing on Rodionenko et al.'s (arXiv:2606.03950) discovery of simultaneous dipolar and quadrupolar topological phases in multi-orbital systems. They demonstrate how 2D systems can support doubly-protected corner states through nested Wilson loops.
A briefing on Fldzhyan et al.'s (arXiv:2606.02451) breakthrough in photonic computing. They demonstrate how Matrix-Vector Multiplication can be performed with a 10x reduction in phase-shift range, significantly reducing thermal crosstalk and power consumption.
A briefing on Wang et al.'s (arXiv:2605.29536) study of single-photon scattering in directional waveguide arrays. They reveal how chiral configurations transform spatial coordinates into time-like evolution, mapping to (1+1)D and (2+0)D Dirac equations.
A briefing on Wang, Hiltunen, and Schotland's arXiv:2605.29536 paper: single-photon scattering in arrays of one-way waveguides, where chiral geometry produces light-cone transport and antichiral geometry restores classical-optics-like interference.
A briefing on Qiu et al.'s experimental realization of an orbital altermagnetic photonic crystal (arXiv:2605.28656). They demonstrate how antiunitary symmetry can enforce momentum-dependent splitting in bosonic systems, providing a physical substrate for complex information routing.
An exploration of deploying a high-fidelity metagenomic pipeline in a home environment using Oxford Nanopore sequencing and GPU-accelerated bioinformatics on the NVIDIA DGX Spark.
A briefing of Hong & Schotland's (arXiv:2605.22797) first-principles QFT framework for single-photon propagation. They establish a universal fidelity formula across erasure and dephasing channels, and discover a fundamental fidelity floor of 1/2 in the strong coupling limit—critical bounds for the next generation of quantum routers.
We followed Apple's four-constraint recipe to the letter, converted a transformer to MLPackage, and got 100% static residency on the first pass. Then we asked the CoreML runtime where it actually ran the ops. 0% on the ANE. Same recipe, scaled to ESM2-class size, scheduler flipped to 100% ANE placement. The post explains why eligibility and placement are different problems, and ships the Swift MLComputePlan tool that distinguishes them without opening Xcode.
We tried to run a 1.1 GB vision model in the user's browser. Most users have 16 GB of RAM and four other tabs. This is what worked (Mol* protein rendering, SmolVLM in VR), what broke (Moondream2 as the default, the Cache API on Safari), and the rule we use now: render on the client, infer on the server, in-browser inference only when latency is the constraint and the model fits in ~500 MB. With the OPFS shim and the commit that ripped the WebGPU inference stack.
PSP measures the Sun directly; JWST observes other bodies that reflect the Sun. Cross-correlate the two and you can measure — and forecast — the irradiance delivered to specific bodies. First end-to-end pipeline live: 6 cloud workflows, public HF dataset, dashboard at ask-meridian.uk/helio. Honest about the v0.1 data-scale gap.
Read Rodney Loudon's Quantum Theory of Light end-to-end to mine ideas for photon-route + meridian. Ten experiments grounded in chapters 3, 5, 6, and 9. Eight came back negative or decorative — squeezing was decorative, photon-number metric strictly weaker than Gaussian fidelity, non-Gaussian heralded states didn't survive the eval scale. One — a g(1) coherence-time feature — shipped to meridian-orbital@3.1.0 with +108% discrimination over what it replaced. Honest accounting with every number.
I tried to make Grok run my pharmacy errands end-to-end. The plan: search → cart → automated checkout, all chat-driven. The wall: VTEX's reCAPTCHA Enterprise enforces hostname binding server-side — your worker forwards a real captcha token, Google reports the origin doesn't match the site key, VTEX silently drops the email and returns 200 {}. The pivot: anonymous cart in-worker + a /checkout/cart/add URL hand-off to the user's browser. Ships ~95% of the value without a Browserbase-class headless browser.
A single-tenant finance MCP that lets Grok read my Binance balance from a country Binance geo-blocks. Three layers, each solving exactly one problem: WebAuthn at the door (passkey-bootstrapped OAuth, single-user), Cloudflare Workers in the middle (auth + tools + KV), a tiny Fly proxy through Bright Data's static-residential AR IP at the edge so Binance's whitelist hears from one fixed IP forever. ~$16/month all in.
A skill-routing classifier had a planet bias — 17 of 18 panel skills got classified as planet. We tried two textbook physics frameworks (Vallado's CRTBP, Sears & Zemansky's spectral classification), neither beat heuristics. Retuned heuristics hit 50% on the panel, then 81% [95% Wilson CI 60%, 92%] on real labelled data. Then we wired an online-SGD loop on top so the classifier improves from every user click — with constant per-request cost and zero local training.
Grok's custom MCP connectors take a server URL + an OAuth dance. We just wired Meridian's orbital skill router into it. The architecture, the OAuth values to paste, and the design choice that lets Grok users authorize with one click — no GitHub PAT, no jargon, no setup wizard.
We just shipped meridian-mcp 2.0.0 to both registries from a single tag push. The honest tradeoff: anonymous install still requires npm, but for everything else GitHub Packages has earned its slot — auth shared with your code, automatic provenance, free private packages, same registry as your Docker images. The full release workflow + the npm pkg set name rename trick that lets one tarball ship under two names.
Most MCP tutorials show npx create-mcp and a Hello-world tool. Real servers in production need auth, per-key quotas, billing, and an upgrade path. Here's the architecture we use for ask-meridian.uk — entirely on Cloudflare's free tier — and the specific gotchas you'll hit.
Most RAG systems either retrieve from a static corpus or end-to-end LLM-rank. Meridian does neither. Llama-3.3-70B authors fresh candidates per task; a deterministic JS classifier assigns each a celestial body class (planet/moon/trojan/asteroid/comet/irregular) based on physics-style features. Why the metaphor, why the deterministic step, and where it goes next.
Workers AI for embeddings (free tier), Vectorize for semantic search (free tier), AI Gateway for caching + observability (free). The whole stack costs $0 until you cross 100k requests/month. Here's the wiring we use for meridian's RAG layer.