I work best in the gaps between content, design, media, tools, apps and operations: shaping awkward workflows into websites, utilities and working systems that people can actually use.
range
A long career in the awkward space between creative and technical work.
web & systems
I make digital life easier.
I have worked across WordPress, hosting, newsletters, video workflows, image handling, analytics, troubleshooting and internal tools, usually in places where the technical, editorial and operational work all overlap.
My strength is seeing the whole system: what the editor needs, what the audience sees, what the CMS allows, what the server is doing and where the process wastes everyone’s time.
I like taking inherited digital machinery and making it clearer, faster, more reliable and less dependent on workarounds.
AI approach
I use AI best when it has a narrow job.
A couple of years ago I started using AI the obvious way: asking it to do whole pieces of work for me. It could be useful, but it was also far too good at producing something that looked finished before it was actually right.
That changed how I use it. I get better results when the work is broken into smaller jobs: one step to frame the task, one to produce something, one to check it, and a human still making the final call. The point is not to make AI sound grander than it is. It is to make it behave more like a tool and less like a very confident intern with no sense of consequences.
That is the thinking behind the diagram. It is not a theory of everything. It is just a practical way to keep intent, production and review from blurring together. The more important the output is, the more I want separate checks: specs, tests, source material, mechanical validation or plain human judgment.
One of the first useful lessons I learned was that AI was often better at building the tool for a job than doing the job directly. If accuracy mattered, I did not want a model improvising the answer every time. I wanted it to help create a repeatable process: calculate the figures, apply the rules, format the output and produce the same result every time. That idea has carried through into small tools, publishing workflows and AI-assisted editorial systems. The useful work is often unglamorous: sorting incoming material, building scaffolds, checking formats, transforming content, producing first drafts and catching boring errors before they become public ones.
I do not see AI as a replacement for taste, context or judgment. Those are usually the valuable parts. I see it as a way to move faster through the repetitive structure around the work, while keeping a person responsible for what matters.
I have 20+ years of video experience, from transferring MiniDV and Betacam SP material for Barbara Corcoran’s NBC segments to filming interviews and market coverage in places like Cannes for World Screen.
That work grew into a full production workflow: shooting, editing, encoding, publishing, archiving and cutting promotional material under real deadlines. At World Screen, I produced roughly 1,500 video interviews, event reports and promotional videos tied to the international TV business.
That work gave me a practical understanding of the whole chain: preparation, capture, lighting, framing, audio, editing, compression, formats, playback and delivery. You learn quickly that video is rarely made in perfect conditions. It is bad rooms, backlight, background noise, difficult color, mismatched cameras, awkward codecs and broken playback, then finding a way to turn imperfect material into something usable and publishable.
Outside my day job, that interest has kept widening. I started shooting more cinematic video in 2008, when the Canon 5D Mark II made shallow depth of field and large-sensor video suddenly feel accessible. Since then I have kept experimenting with composition, light, lenses, frame rate, depth of field and the settings that shape how an image feels, while also getting pulled into less standard media: VR180, 360 video, spatial video, 3D, projection, upscaling, interpolation, immersive playback and self-hosted media systems. I am interested in both sides: the image itself and the system that gets it to someone’s screen.
photography
Photography trained my eye.
I got my first SLR when I was 15 and started reading photography books before I really knew what I was doing. The interest was always both technical and visual: cameras, lenses, light, composition, color, focal length, exposure and the small choices that change how an image feels.
Getting a DSLR in 2005 opened the floodgates. I could shoot constantly, work in RAW, experiment properly and learn from the feedback loop. Since then I have done a lot of client photography, mostly portraits, along with retouching, image selection, two weddings and plenty of practical work making images polished, usable and appropriate for the job.
Photography trained a useful reflex: looking at a scene and immediately seeing the trade-offs. Light direction, background, distortion, crop, expression, depth of field, noise, sharpness, color and where the eye goes first. I use Lightroom and Photoshop, mostly Lightroom, and I still follow camera technology closely, from sensors and colour science to phone cameras and computational photography.
I like the nerdy machinery of image-making, but I also like the simple pleasures: faces, atmosphere, golden hour, pinks and yellows, stereo photography and the moment when an ordinary scene becomes worth keeping.
writing
I can actually string a sentence together.
A shocking amount of technical friction comes down to the fact that nobody wrote down clearly what the thing is supposed to do. I fix that.
From punchy product copy to plain-English documentation for executives, I translate complex systems into words that human beings can actually tolerate reading.
newsletters
I sent 300 million marketing and news emails
First, my apologies. Second, I know mass email a little too well. Newsletter construction, delivery quirks, deliverability and landmines.
The work covered the whole unpleasant chain: HTML construction, inline styling, image handling, link tracking, list segmentation, subject lines, advertiser materials, test sends, rendering problems, delivery quirks and the strange little disasters that appear five minutes before send.
What I learned is that email is less about elegance than reliability. You need systems that are repeatable, hard to break and clear enough that small mistakes do not become public ones. I know how to keep campaigns moving, how to spot trouble early and, more importantly, what usually breaks.
current projects
Live sites, tools and experiments that grew out of actual problems.
MODERN STUDIO REPORT
A practical guide to the new production technology reshaping the TV business
Live publicationWordPress editorial platformAI-assisted workflow
Modern Studio Report is a trade publication I built for the parts of the TV business that are changing fastest: AI, cloud workflows, virtual production, localization, rights, tools and the production stack around them.
After 17 years inside TV trade publishing, I had seen the industry change dramatically: streaming, cloud production, AI tools, virtual production, automated localization, new rights questions, new buyer behavior, new pressure on marketing teams, and a lot of confusion hiding behind confident language. I wanted to build a site that treated those shifts seriously without sounding like a technical brochure or another breathless AI newsletter.
It is written for people who need to understand this stuff without becoming specialists: studio executives, production teams, distributors, marketing departments and service companies. The tone is plain English, but not dumbed down.
WHAT THE SITE CONTAINS
A structure built for more than daily news.
01
News
Fast-moving coverage of production technology, business shifts and industry change.
02
Explainers
Plain-English context for complex tools, workflows and technical terms.
03
Ecosystem
Company and platform profiles that map the people building the new stack.
04
Case Studies
Structured evidence around what changed, what worked and what failed.
05
Tools
Practical decision aids for evaluating production technology.
EDITORIAL MODEL
The site runs on WordPress with a custom theme and a publishing structure designed around the needs of a modern B2B media brand. It includes news, features, explainers, company profiles, case studies and a separate tools section. The editorial model is meant to support both fast-moving coverage and slower reference material, so the site can work as a daily publication and as a useful industry resource over time.
A lot of the work is in the structure. Explainers are designed to give clear context around subjects like AI dubbing, virtual production, synthetic media, cloud workflows and production automation. Company pages help map the ecosystem. Case studies are built to capture the practical side of technology adoption: what problem was being solved, what changed, what worked, what failed and what should be treated with caution. I wanted the site to do more than rewrite launch announcements
HOMEPAGE LOGIC
The homepage is configurable and editorially controlled, with sections for lead stories, briefs, explainers, company highlights and discovery. It can be curated manually or filled automatically, with logic to prevent repeated stories and preserve the layout while the site is still growing. Many publishing layouts fall apart the moment the content mix is imperfect. I wanted the system to handle real-world mess.
SEARCH & DISCOVERY
Search is also treated as part of the experience rather than a default WordPress afterthought. One of the ideas I liked most was having clean editorial blocks transform into search results, so discovery feels like part of the publication rather than a trip to a separate utility page. The homepage browse area, archives and search hub are meant to help users move naturally from reading into finding.
SEARCH BECOMES EDITORIAL
A search result should still feel like a publication.
SearchAI dubbing
BEST EXPLAINERAI dubbing workflow
A plain-English guide to automated localization, voice matching and approval risks.
RELEVANT COMPANYCloud localization platform
Company profile · Source indexed · Production tools
PRACTICAL TOOLRights readiness checklist
Tool · Updated recently · Questions before adoption
Search becomes a guided editorial surface, not a generic results page.
DESIGN SYSTEM
The design has two related but distinct modes rather than a simple light/dark color swap. Light mode is airier, more open and a little more newspapery. Dark mode leans further into a sleek, app-like interface, with more contained panels and a stronger sense of depth. That decision came from building several projects and realizing that theme switching is a wasted opportunity if it only changes the background color. People who prefer dark mode often seem to want a different mood, not just the same site inverted.
TWO MODES, TWO MOODS
Same site, different atmosphere.
There are also many smaller details that may go unnoticed, but still shape the experience: careful card layouts, image handling options, logo fallbacks, article deks, further-reading links, cleaner archive filters, useful empty states, theme-aware interface elements and admin conveniences to reduce repetitive publishing work. Beneath that surface is a deeper layer of custom structure, both front end and back end, built to make WordPress behave less like a generic theme and more like a tailored editorial platform. I wanted the public site, the publishing workflow and the back-end logic to feel like one system.
Incoming feeds→Signal Room→Human judgment→Draft→WordPress→Modern Studio Report
The publishing workflow extends beyond WordPress. I also built an external editorial dashboard, The Signal Room, to help collect, score and sort incoming industry stories from feeds and other sources. It supports AI-assisted drafting, but with human judgment still in charge. Modern Studio Report is the public expression of that wider system: a site, a workflow and an editorial engine designed together.
The project pulls together most of the work I know how to do: writing, editing, design, WordPress development, information architecture, automation, AI workflow design, search and publishing operations and the grim little details that make systems either pleasant or unbearable to use. It involves writing, editing, design, WordPress development, information architecture, automation, AI workflow design, search and publishing operations. It has also been a useful test: what I can build when I am not just maintaining an existing system, but deciding what the system should be.
Under the hood: Built with an algorithmic homepage fallback, a dedicated
high-relevance Ajax-backed search hub, and intelligent authoring conveniences like safe SVG
ingestion and remote title resolution.
THE SIGNAL ROOM
A local editorial cockpit for finding usable stories in the news firehose.
The Signal Room is a local-first editorial cockpit I built for Modern Studio Report: a private workspace for collecting incoming industry material, ranking it, grouping duplicates and deciding what deserves attention before anything becomes a draft.
It is built around one of the most tedious parts of trade publishing: reading dozens of press releases, working out which ones matter and rewriting the useful ones into short news items that often add little beyond making the release usable. Feed, email and social ingest are designed to take on more of that first-pass grind, freeing the editor for work that actually benefits from a human being: interviews, reported pieces, analysis and deeper features.
It pulls the noisy parts of trade publishing into one place: feeds, source lists, emails, social ingest paths, story clusters, AI review queues, draft generation, model settings and WordPress handoff. The working principle is simple: cheap deterministic signal first, optional AI second, human judgment always in charge. When AI is used, it can also research around an item and provide context, so the draft is not just a laundered press release with nicer shoes.
Before any model is asked to think, the system uses explainable scoring, source weighting and story grouping to reduce the mess. AI is used where it helps with judgment or drafting, not sprayed over the product because AI is fashionable.
The system is built around task-based model selection rather than one fixed AI provider. Item review, story matching, social triage, drafting and research-assisted writing can each be pointed at the model or provider that makes sense for the job: local when privacy, cost or speed matter; cloud when web grounding or stronger drafting is worth the trade-off.
Item review
Relevance, action, confidence
Story matching
Clusters, fingerprints, same-story checks
Social triage
Keyword gate, optional LLM
Drafting
Local or cloud drafting modes
Research-assisted writing
Search-grounded context
TASK ROUTERPer-task model choice
Local model
Private, cheap, controllable
OpenAI-compatible provider
LM Studio or similar endpoints
Gemini / grounded web
Grounded drafting option
OpenAI web search
Web-search drafting option
Brave research snapshot
Research before drafting
WordPress draft handoff
Human approval stays required
The workflow can route different jobs to different models: local when privacy, speed or cost matter; cloud when research or stronger drafting is worth it.
PRODUCT VIEWS
Screens from the working tool.
The Signal Room
Watchlist overviewAI review detailDraft workspace
BUILT AS A WORKING LOCAL SYSTEM
Practical machinery, not a landing-page sketch.
INGEST
Feed sources
Email worker
Social ingest paths
Local cache
TRIAGE
Deterministic scoring
Story clustering
Optional AI review
Human decisions
OUTPUT
Draft generation modes
WordPress REST handoff
WordPress drafts by default
565+ tests
Built as a single-user local editorial system, not an enterprise newsroom platform. The point is practical leverage: one editor seeing more signal, wasting less time and keeping the final decision human.
Under the hood: React + Vite, Express, SQLite, deterministic scoring before AI review, story-group deduplication, local model routing, cloud drafting options and WordPress REST handoff.
DIPTYCH
The best part of Dreamweaver, rebuilt: open an ordinary HTML file, edit it by
sight, and save it without the editor mangling your code.
Local-first desktop appTauri + Rust shellVite / vanilla JSSource-preservingNo DOM serialization on save
A two-pane editor — Design on the left, Source on the right — kept
perfectly in sync.
WYSIWYG editors are all but gone from the landscape. Design view is an afterthought in Dreamweaver's most recent version. The entire app is all but abandoned and locked behind an expensive subscription model. Diptych recovers the one thing Dreamweaver was great for: opening a normal HTML file,
seeing it visually, editing it directly, and saving it without the editor rewriting or damaging the
original code. It is deliberately narrow — not a website builder, CMS, or publishing platform. Just a
lightweight tool for editing local HTML files with a live relationship between Design and Source.
Source Preservation
The hard part wasn’t rendering HTML — it was preserving it.
Most visual editors save by serializing the browser’s DOM, which quietly reorders attributes,
normalizes whitespace, and strips the odd-but-intentional code you hand-wrote. Diptych treats the
original source buffer as the authority. Design view is only a surface; when you edit, it maps the
change back to the exact source range and patches only that. Untouched source stays
untouched. It can even open static accordions, tabs, and <details> in Design view
— without running their scripts or saving any temporary UI state — so hidden content can be edited
safely.
Never serializes the DOM to saveNever runs the page’s scriptsPatches only the ranges you edit
Under the hood: A Tauri desktop app — Rust/Tauri shell, Vite + vanilla-JS frontend
— that opens files from disk and writes them back. The source buffer is the single source of truth;
Design view is a render-and-edit surface only. No DOM or Shadow-DOM serialization on save, and no
page scripts are ever executed. AI-assisted to build, but the scope, constraints, source-preservation
rules, and regression tests were mine.
Useful little instruments for digital odds and ends.
A recent afternoon with my Kindle Scribe led to an oddly useful idea: when I wasn’t using it, it would make a lovely picture frame. The catch was that the only practical way to do it was to turn an image into a tiny EPUB file, which was much more fiddly than it needed to be. So I built a one-step tool for it, then put it online in case anyone else wanted the same thing.
That small utility became a testbed for sharpening my AI-assisted coding workflow. I gave myself a deliberately unreasonable deadline: build and launch a suite of 12 tools in a couple of days. The tools and the site went live the same day. It has since grown into a live collection of 60+ utilities for common digital chores: some silly, some genuinely useful.
I built it entirely client-side, meaning zero data leaves your browser. No accounts, no tracking. Just fast, good-looking tools that do exactly what they say.
Modern Studio Report taught me that a good dark mode is a different mood, not just an inverted background. Here I pushed it further. Pop is loud, sunny, confident and retro in an over the top not-too-distant-past kind of way. Nova is calm, dark and quiet. The switch up top isn’t a screenshot — it re-skins this very panel, so someone who loves bold color and someone who wants the lights down can both feel at home.
AI testbed
A testbed for turning AI help into working tools.
I also used the site as a testbed for AI-assisted development. I tried several approaches, including running agents in parallel, to build tools out faster. It also made the limits obvious: models help most when the job is tightly scoped, the rules are written down and the output can be checked.
So the site isn’t hand-typed and it isn’t blindly generated. I built a small production line around the work — a queue, a scaffold generator, and a roughly 500-line mechanical verifier with an agent-readable spec — so anything an agent produces has to pass real checks before it counts as a finished tool.
Kindle Wallpaper MakerThe one that started it — frame an image and package it as a Kindle cover EPUB.PDF Slicer & JoinerInspect, pull and reassemble PDF pages in-browser with PDF.js.Image CropperCanvas crop UI with drag handles, a dark mask and a rule-of-thirds grid.Privacy BlurPaint over names and numbers in a screenshot, then export it redacted.Audio SnipperTrim the start and end of a clip without opening a desktop editor.Palette PullerPull the dominant colors out of any image as copyable hex.
Under the hood: Eleventy static-site build, vanilla JS (no framework), a ~3,000-line token-driven dual-theme CSS system, a shared ~53 KB browser library (Canvas, PDF.js, JSZip, Web Audio), a scaffold generator plus a ~500-line mechanical verifier, and an agent-ready tool-building spec. Ships as static files.
SPATIAL SILO
One canonical library for immersive media: VR180, 360, spatial video, 3D,
flat video and images, organized by metadata instead of folder compromises.
Rust + AxumSQLite asset indexFFmpeg media jobsQuest/WebXR playback2D + 3D workflows
2D browser playbackWebXR routeAny-device library modelExperimental ML pipelines
Spatial Silo is my most technically ambitious personal project. The
idea is a self-hosted media library for the material normal apps handle badly: VR180 clips, 360
video, Apple Spatial Video, SBS/TB 3D, flat video, images and camera-native immersive formats.
Instead of splitting everything into separate libraries, Spatial Silo treats format, camera, tags and collections as metadata. Everything
enters one asset pool, then cameras, formats, tags, smart collections and manual collections become
metadata-driven views over the same library. The Plex model has you create different libraries for different media. I wanted a facet-first approach that handles the stranger
end of personal media, and with enough structure that the library can eventually work across ordinary
screens, 3D displays, headsets and browser-based playback.
Projection aware playback
2D, 3D, 180, 360, spatial and derived media routes
01Browser playback
Inline 2D video, image viewing and asset detail pages for ordinary screens.
02Immersive preview
WebXR and A-Frame paths for VR180, 360 and stereo media, with projection overrides.
03Quest-oriented route
Hardware media layers via XRMediaBinding / XREquirectLayer for high-resolution playback.
ADVANCED MEDIA PIPELINES
Beyond cataloging, I wanted room for media-improvement pipelines: interpolation, derived formats and experimental 2D-to-3D work.
60fps interpolation
Temporary playback or permanent library export.
FFmpeg minterpolate→RIFE AI modes→Progressive HLS→Save as asset
Fast frame blending, higher-quality optical-flow/MCI, RIFE via rife-ncnn-vulkan, smart
buffering, cancellation, cleanup and export-to-library jobs.
2D to 3D generation
Experimental depth-based stereo generation.
Live 3D HLS sessions, low-resolution depth sidecars, ONNX/CoreML depth inference, DIBR stereo
synthesis and offline SBS export paths. It is a research pipeline inside the product, not a
finished consumer feature yet.
Under the hood: Rust/Axum backend, SQLite metadata index, watched-source scanner,
FFmpeg thumbnails and HLS derivatives, ffprobe + ExifTool analysis, vanilla JavaScript frontend,
WebXR playback paths, background job tracking, RIFE interpolation, and experimental depth-based 3D
generation.
track record
When people hire me, they tend to keep me around.
“
Ricardo Guise
President & Publisher, World Screen
He's exceptionally good at what he does. But what I'd also want any prospective employer to know is that Simon is a genuinely interesting person to have around. 2026
March 25, 2026
To Whom It May Concern,
I’m writing to recommend Simon Weaver, who joined us in 2008 and served as our Online Director right up until our closure this year. That’s nearly two decades of dedication, and in that time, he became the backbone of everything we did online.
Simon was responsible for the full scope of our digital operations—from developing and maintaining our many newsletters and websites to designing and editing our virtual video festivals and other online presentations. He was also responsible for the delivery of all our daily and weekly newsletters, as well as our promotional material. He provided clients with digital impact reports, giving them detailed stats on their campaigns. As the company moved increasingly toward digital publishing; his role only grew. He took that transition in stride and kept everything running smoothly, even as the demands on him kept expanding.
He’s exceptionally good at what he does. But what I’d also want any prospective employer to know is that Simon is a genuinely interesting person to have around. He brings real energy to the people he works with, and I’m going to miss our conversations more than I can say.
We’re closing after forty years, and Simon leaving is one of the harder parts of that. I’d recommend him without reservation for any organization looking for someone who combines serious technical skill with the kind of personality that makes a workplace better.
Sincerely, Ricardo Guise President & Publisher World Screen
“
Barbara Corcoran
Founder, The Corcoran Group · Shark Tank investor
"Simon is a real gem! He is my web + tech master, video master and writer extraordinaire!" 2008.
a little background
Englishman in Brooklyn.
A few detours, a lot of making things up as I went along, and a long-running preference for making complicated things clearer.
I’m from Rochdale, in the north of England, and I’ve lived in Brooklyn since 2003.
Before my run in TV trade publishing, I had a mixed working life: factory summers, supermarkets, a nursing home, country fairs, British Gas billing problems, cheffing on the Isles of Scilly, English teaching in Granada and university teaching in Seoul.
Working with Barbara Corcoran added another kind of training: writing TV pitches, meeting executives, helping turn loose ideas into segments and seeing the work that sits behind something that looks simple on screen.
That range has been useful. I’m comfortable moving between different kinds of people, different kinds of work and problems that do not belong clearly to one department. It is probably why I ended up being useful in the spaces between roles.
PRIOR STOPS: FACTORY · KITCHEN · CLASSROOM · MEDIA
I studied Philosophy and English Literature at university. I thought I’d cleverly merge the two and do a master’s in Critical Theory. I quit at 3 a.m. one night in the University of Nottingham’s computer lab.
I was trying to wrangle Baudrillard and Lacan into a thesis on God only knows what. Then came the epiphany: the words on my page were gibberish. The experience put me off academia for life.
That’s the end of my philosophy story. I at least learned that making simple ideas complicated wasn’t a good use of my time, or anybody else’s. I prefer it the other way around.
BaudrillardLacanCritical TheoryPlain English
I’m interested in useful digital work: websites, publishing systems, editorial tools, AI-assisted workflows, video and media projects, and the awkward problems that appear when one person needs to cover several departments at once.
Ricardo Guise
