Carl Welch
Work
Contact →

Work / 2026

Selected work

8 detailed case studies and 12 selected public repositories spanning product systems, applied AI, audio software, and creative technology.

Selected work / 01

Job Search OS

Applied AI / Product systemsProtected single-user production appProject index
Job Search OS dashboard showing the daily review cockpit and application pipeline
Application assistantSearch operationsAgent review boardResume workspaceSearch profilesCompany sources

A private operating system for discovering roles, reviewing fit, assembling evidence-backed application materials, and tracking the work around a job search.

  • Next.js
  • TypeScript
  • LangGraph
  • LangSmith
  • PostgreSQL + pgvector
  • OpenAI
  • MCP
View Job Search OS
Human-reviewed career operations13 components · 13 connections

Four product surfaces share one evidence and orchestration layer; every consequential external action stops at an approval boundary.

Product surfacesWays Carl enters and reviews work
Application + orchestrationServer-owned workflow boundary
Durable stateShared operational record
Controlled exitsNo autonomous submission
  1. Next.js · React
    Daily cockpit

    Search, fit, materials, review

  2. Chrome extension
    Browser capture

    Role and source intake

  3. Slack
    Jolene command

    Private operator channel

  4. MCP · stdio
    Local tool access

    Tracking and preparation tools

  5. Next.js App Router
    Workflow API

    Typed routes and policy checks

  6. OpenAI · LangSmith
    Specialist agents

    Structured outputs + redacted traces

  7. pgvector
    Evidence RAG

    Approved career evidence only

  8. LangGraph
    Durable workflows

    Postgres checkpointing + review state

  9. Node.js · Redis
    Background worker

    Embeddings and queued work

  10. PostgreSQL
    System of record

    Roles, claims, runs, materials

  11. pgvector
    Semantic index

    Career evidence embeddings

  12. Policy gate
    Owner approval

    Review before consequence

  13. Manual / approved
    External channels

    ATS, email, LinkedIn

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Daily cockpit → Workflow API
  • Browser capture → Workflow APIcapture
  • Jolene command → Specialist agents
  • Local tool access → Workflow API
  • Workflow API → Specialist agentsplan
  • Workflow API → Evidence RAG
  • Specialist agents → Durable workflows
  • Evidence RAG → Semantic index
  • Durable workflows → System of record
  • Durable workflows → Background worker
  • Background worker → System of record
  • Specialist agents → Owner approvalpropose
  • Owner approval → External channelsapproved only

Selected work / 02

Flight Tracker AI

Geospatial / Aviation intelligenceDeployed read-only portfolio demoProject index
Flight Tracker AI live San Francisco traffic workspace showing 158 aircraft, weather layers, a selected aircraft, and the regional flight list
Deterministic replay workspaceRoute comparisonMobile replay

A browser-based aviation intelligence system that brings live or replayed traffic, weather, hazards, trajectories, and explainable attention cues into one map-led interface.

  • Next.js
  • TypeScript
  • Rust + Axum
  • PostgreSQL + PostGIS
  • NOAA data
  • Vercel
View Flight Tracker AI
Regional aviation intelligence12 components · 13 connections

Live providers and deterministic fixtures enter one Rust domain service; the browser receives a sanitized, read-only air picture.

External observationsBest-effort public data
Domain serviceRust owns aviation policy
Hosted platformRead-only public deployment
  1. ADSB.lol · Airplanes.live
    Aircraft positions

    Primary + fallback feeds

  2. NOAA AWC
    Aviation weather

    METAR, TAF, PIREP, SIGMET

  3. nowCOAST · GFS/HRRR
    Weather imagery

    Radar and forecast layers

  4. OpenFreeMap · OSM
    Base map

    Vector tiles and geography

  5. Rust · Axum
    Ingestion + normalize

    Provider fallback and typed records

  6. Rust · PostGIS
    Spatial engine

    Trails, regions, route geometry

  7. Deterministic Rust
    Attention policy

    Explainable rules and factors

  8. OpenAI + fallback
    Constrained wording

    Synthetic recommendation phrasing only

  9. Neon · PostgreSQL/PostGIS
    Spatial record

    Aircraft, observations, scenarios

  10. Deterministic JSON
    Replay fixtures

    Repeatable review scenarios

  11. Render · Axum
    Public read API

    Sanitized, no-store responses

  12. Vercel · Next.js · MapLibre
    Interactive airspace

    Traffic, weather, replay, comparison

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Aircraft positions → Ingestion + normalize
  • Aviation weather → Ingestion + normalize
  • Weather imagery → Interactive airspace
  • Base map → Interactive airspace
  • Ingestion + normalize → Spatial record
  • Ingestion + normalize → Spatial engine
  • Spatial record ↔ Spatial engine
  • Replay fixtures → Spatial enginereplay
  • Spatial engine → Attention policy
  • Attention policy → Constrained wording
  • Attention policy → Public read API
  • Constrained wording → Public read API
  • Public read API → Interactive airspace

Selected work / 03

Wave Factory Essentials

Audio software / DSPPre-release tester buildsProject index
Threefold Palm audio plug-in interface from the Wave Factory Essentials collection
Valley SpiritThreefold Palm / gesture stateValley Spirit / gesture stateThreefold Palm / Temple of MasterySpirit Mirror / visual foundation

A family of focused audio plug-ins designed as real AU, VST3, and CLAP products, with custom interfaces, local signal processing, and host-level validation gates.

  • C++
  • iPlug2
  • AU
  • VST3
  • CLAP
  • DSP
View Wave Factory Essentials
Host-loadable real-time audio system12 components · 12 connections

The DAW, plug-in adapter, real-time DSP, interface, and optional camera controller remain deliberately separated by thread and process boundaries.

Audio hostMusician-owned runtime
Plug-in processHard real-time boundary
Optional local controlNever required for audio
Release gatesSeparate evidence, not one checkbox
  1. Logic · host DAW
    DAW session

    Audio, automation, saved state

  2. AU · VST3 · CLAP
    Format adapters

    Host lifecycle and parameter bridge

  3. iPlug2 parameters
    Host automation

    Stable IDs and normalized values

  4. C++ · iPlug2
    iPlug2 adapter

    Host events into product code

  5. wave_factory_dsp
    Shared DSP core

    No allocation, locks, file or network I/O

  6. C++ real-time audio
    Product signal path

    Split, shape, delay, diffuse, mix

  7. Lock-free state
    Meter bridge

    Relaxed atomics leave audio thread clean

  8. iPlug2 UI
    Custom interface

    Controls, help, metering, art direction

  9. macOS helper
    Camera helper

    Local capture outside plug-in process

  10. Apple Vision
    Hand pose model

    Two 21-point hand skeletons

  11. CC 20–25 · local IPC
    Gesture bridge

    Six signals over local bridge or MIDI

  12. auval · CI · notarization
    Validation pipeline

    Host · listening · signing · packaging

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • DAW session → Format adapters
  • Format adapters → iPlug2 adapter
  • Host automation → iPlug2 adapter
  • iPlug2 adapter → Shared DSP core
  • Shared DSP core ↔ Product signal path
  • Product signal path → Meter bridge
  • Meter bridge → Custom interface
  • Custom interface ↔ iPlug2 adapter
  • Camera helper → Hand pose model
  • Hand pose model → Gesture bridge
  • Gesture bridge → Host automationoptional
  • iPlug2 adapter → Validation pipelinebuild artifacts

Selected work / 04

Supraconscious Avatar AI

AI product / Reflection systemsDeployed web application / mobile scaffoldProject index
Authenticated Supraconscious journal workspace with a private reflection draft, response preferences, and Guide response
Reflection historyConstant GuidePrivacy and preferences

A multi-surface reflection product that turns journal entries into guided questions while giving the operator a separate place to govern prompts, sources, safety, and review.

  • Next.js
  • TypeScript
  • PostgreSQL + Prisma
  • OpenAI
  • MCP
  • Flutter
View Supraconscious Avatar AI
Governed reflection platform13 components · 14 connections

Customer surfaces share a policy-first AI package and database, while a separately deployed admin plane controls prompts, sources, safety, and feature state.

Experience surfacesUser and compatible AI clients
Shared AI runtimePolicy before generation
Durable knowledge + stateServer-only persistence
Operator control planeSeparately deployed admin
  1. Next.js
    Journal web

    Writing, reflection, privacy controls

  2. Express · MCP
    ChatGPT surface

    Authenticated tools + widget

  3. Flutter · scaffold
    Mobile client

    Scaffolded product surface

  4. Next.js route handlers
    Server routes

    Auth, journal, voice, billing

  5. OpenAI · Zod
    Safety classifier

    Intensity and response boundaries

  6. Shared AI package
    Reflection guide

    One active voice + structured output

  7. Policy-first keyword RAG
    Governed retrieval

    Eligible reviewed sources + citations

  8. GraphRAG · feature-flagged
    Ontology context

    Approved neighborhoods only

  9. Shared AI package
    Pattern memory

    Traces and reusable reflection state

  10. PostgreSQL · Prisma
    Application record

    Journal, sources, prompts, traces

  11. PostgreSQL
    Reviewed knowledge

    Chunks, provenance, ontology state

  12. Separate Next.js app
    Admin console

    Prompts, sources, safety, quality

  13. Stripe
    Subscription boundary

    Plans and billing operations

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Journal web → Server routes
  • ChatGPT surface → Reflection guide
  • Mobile client → Server routesscaffold
  • Server routes → Safety classifier
  • Safety classifier → Reflection guide
  • Reflection guide ↔ Governed retrieval
  • Governed retrieval → Reviewed knowledge
  • Ontology context → Reviewed knowledgeflagged
  • Reflection guide → Pattern memory
  • Pattern memory → Application record
  • Server routes → Application record
  • Admin console ↔ Application record
  • Admin console ↔ Reviewed knowledge
  • Server routes → Subscription boundary

Selected work / 05

Argent Matchmaking

Product strategy / Private servicesSynthetic concept prototypeProject index
Argent Matchmaking product system showing design tokens, public web, mobile introduction, and matchmaker review workspace
Relationship directionService environmentNocturne image system

A human-led matchmaking concept spanning a discreet public experience, member mobile surface, matchmaker workspace, shared design system, and the beginnings of a production-minded platform.

  • Next.js
  • TypeScript
  • Fastify
  • Flutter
  • PostgreSQL
  • Docker
View Argent Matchmaking
Private-service product foundation12 components · 16 connections

The current synthetic concept is split into public, staff, mobile, API, worker, contract, and server-only policy boundaries without implying a live matching service.

Cross-platform surfacesSynthetic experience layer
Service + contract boundaryVersioned and generated
Server-only foundationNo real profiles or matching
  1. Next.js
    Public experience

    Editorial concept and application story

  2. Next.js · separate app
    Staff workspace

    Separate matchmaker review surface

  3. Flutter
    Member mobile

    Cross-platform introduction concept

  4. OpenAPI · TypeScript · Dart
    Generated clients

    One contract for web and mobile

  5. Fastify
    HTTP service

    Versioned workflow boundary

  6. Web + Flutter tokens
    Nocturne system

    Shared semantic tokens and adapters

  7. Framework-light TypeScript
    Domain policy

    Server-only rules and calculations

  8. Node.js worker
    Background worker

    Process and delivery foundation

  9. PostgreSQL 18
    Data foundation

    Migrations and synthetic fixtures

  10. Redis 8
    Queue foundation

    Local background coordination

  11. Docker Compose
    Isolated runtime

    Five-service smoke verification

  12. GitHub Actions
    Supply-chain gates

    Secrets, scanning, SBOM, containers

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Public experience → Generated clients
  • Staff workspace → Generated clients
  • Member mobile → Generated clients
  • Nocturne system → Public experience
  • Nocturne system → Staff workspace
  • Nocturne system → Member mobile
  • Generated clients → HTTP service
  • HTTP service → Domain policy
  • HTTP service → Data foundation
  • HTTP service → Queue foundation
  • Queue foundation ↔ Background worker
  • Background worker → Data foundation
  • Isolated runtime → HTTP service
  • Isolated runtime → Background worker
  • Isolated runtime → Data foundation
  • Supply-chain gates → Isolated runtimeverify

Selected work / 06

Jolene AI

AI agents / Knowledge systemsDeployed career-wide public delegateProject index
Jolene AI character portrait used for Carl Welch’s personal chief-of-staff agent
Nevada field campConversation state ensembleCanonical identity lockAbandoned greeting experimentApproved runtime atlas

A governed AI system that uses public hybrid RAG over reviewed evidence, bounded conversation continuity, tested model and deterministic responses, explicit action boundaries, and a warm, recognizable personality across my full 20-plus-year career.

  • TypeScript
  • OpenAI
  • Hybrid RAG
  • Next.js
  • PixiJS
  • SQLite
  • Docker
  • Slack
  • Vercel
  • Playwright
View Jolene AI
Private chief of staff, career-wide public delegate, and measured release system13 components · 14 connections

Private memory and action tools stay in Carl’s local trust zone. The public path uses hybrid retrieval-augmented generation (RAG) over a reviewed career artifact, bounded continuity, grounded model or deterministic composition, and a separate evaluation and promotion pipeline.

Private JoleneCarl-owned memory and action zone
Public portfolio delegateRead-only career conversation zone
Composition and releaseMeasured quality and promotion gates
  1. TypeScript · Socket Mode
    Owner surfaces

    Local control center and verified Slack

  2. Portable agent runtime
    Private agent core

    Tasks, memory, policy, reviews

  3. SQLite · Obsidian
    Private state

    Memory, audit, evidence, delivery ledger

  4. Human authorization
    Exact action gate

    Intent, recipient, content, expiry

  5. Next.js · React
    Portfolio chat

    Questions, citations, role comparison

  6. Next.js · TypeScript
    Same-origin BFF

    Bounded schemas and server-only token

  7. Vercel · TypeScript
    Public Jolene API

    Answer, job-fit, policy refusal

  8. Lexical + embeddings
    Public hybrid RAG

    Reciprocal-rank fusion over approved evidence

  9. Versioned · hash verified
    Reviewed career artifact

    Five chapters, 16 roles, 92 published records

  10. Typed conversationContext
    Bounded continuity

    Public context only, short lived, resettable

  11. OpenAI · structured output
    Grounded composer

    Model or deterministic answer, then validate

  12. Playwright · eval runner
    Launch evaluation

    132 cases, 192 turns, red team

  13. CI · Vercel
    Release gates

    Preview, corpus pin, promote, rollback

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Owner surfaces ↔ Private agent core
  • Private agent core ↔ Private state
  • Private agent core → Exact action gate
  • Same-origin BFF → Public Jolene API
  • Portfolio chat → Same-origin BFF
  • Public Jolene API → Public hybrid RAG
  • Public hybrid RAG → Reviewed career artifact
  • Public Jolene API ↔ Bounded continuity
  • Public Jolene API → Grounded composergrounded context
  • Launch evaluation → Grounded composerscore
  • Launch evaluation → Reviewed career artifactcoverage
  • Release gates → Public Jolene APIpromote
  • Private state → Reviewed career artifactreviewed export only
  • Release gates → Portfolio chatportfolio pin

Selected work / 07

ProgressionLab

Creative technology / AI music systemsDeployed public web productProject index
ProgressionLab generator with seed chord, mood, mode, genre, style reference, and adventurousness controls
Playable suggestionExamples library

An AI-assisted harmony and songwriting workspace that turns musical intent into playable chord progressions, practical voicings, visual explanations, and exportable session material.

  • Next.js
  • TypeScript
  • OpenAI Responses API
  • Tone.js
  • Prisma + PostgreSQL
  • WebAuthn
View ProgressionLab
Playable AI music workspace12 components · 14 connections

Musician-facing tools share one typed application boundary, while persistence, prompts, access policy, and operational controls remain behind server-owned interfaces.

Creative surfacesHear, inspect, and continue ideas
Application + music logicTyped generation and playback boundary
Persistence + operationsAccess, content, prompts, and telemetry
  1. Next.js · React
    Harmony generator

    Intent, mood, mode, and style

  2. Web MIDI · Canvas
    Chord Explorer

    MIDI input and harmonic context

  3. React · App Router
    Library + sharing

    Saved, public, and curated ideas

  4. Next.js route handlers
    Generation API

    Validation, access, and normalization

  5. OpenAI · strict schema
    Structured suggestions

    Chords, voicings, and structure

  6. Tone.js
    Playback engine

    Tempo-aware chord audition

  7. MIDI · PDF
    Session exports

    Portable musical material

  8. Typed TypeScript
    Music-theory model

    Voicings, diagrams, and inference

  9. Prisma · PostgreSQL
    Product record

    Users, progressions, and prompts

  10. Separate Next.js app
    Operations console

    Prompts, content, pricing, analytics

  11. Sessions · WebAuthn
    Identity + entitlement

    Protected actions and security keys

  12. Sentry · server events
    Operational telemetry

    Errors, usage, and funnel events

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Harmony generator → Generation API
  • Chord Explorer → Music-theory model
  • Library + sharing → Generation API
  • Generation API → Structured suggestions
  • Generation API → Music-theory model
  • Music-theory model → Playback engine
  • Music-theory model → Session exports
  • Generation API → Product record
  • Generation API → Identity + entitlement
  • Operations console → Product record
  • Operations console → Structured suggestionsprompts
  • Library + sharing → Product record
  • Generation API → Operational telemetry
  • Operations console → Operational telemetry

Selected work / 08

EchoAtlas

Geospatial / SAR evidence systemsPublic portfolio demonstrationProject index
EchoAtlas Analyze workspace with public Umbra-derived before and after imagery, machine-generated candidate outlines, and an analyst review queue
Explore provider availabilityReview pair comparabilityAnalyze prepared evidence

A human-in-the-loop synthetic aperture radar workbench for discovering provider-reported imagery, reviewing acquisition pairs, and inspecting deterministic change candidates without turning a heuristic into a claim about the world.

  • Python
  • FastAPI
  • Rasterio + NumPy
  • React + TypeScript
  • MapLibre GL JS
  • Vite
  • Docker
  • Vercel
View EchoAtlas
Navigation, evidence, processing, and interpretation stay separate14 components · 14 connections

The public interface can discover metadata, compare a pair, and open one approved prepared demonstration. Deterministic raster processing remains local, and a versioned evidence bundle connects it to human review.

Human review surfacesNavigate, compare, inspect, assess
Public application boundaryValidated metadata and pair contracts
Deterministic local processingPinned inputs and reproducible artifacts
Evidence and deliveryPortable records and bounded hosting
  1. React · MapLibre
    Explore

    AOI, provider search, footprints

  2. Semantic HTML
    Accessible results

    Map and list selection parity

  3. React dialog
    Pair review

    Comparability before compute

  4. React workbench
    Analyze

    Candidates, evidence, assessments

  5. FastAPI · Pydantic
    Workflow API

    Strict runtime validation

  6. Umbra · Sentinel-1
    Provider adapters

    Normalized partial-result search

  7. JSON · GeoJSON
    Immutable selection

    Pair identity + content hash

  8. Python · AWS CRT
    Pinned acquisition

    Allowlisted, checksum verified

  9. Rasterio · NumPy
    Local raster pipeline

    Crop, align, normalize

  10. SciPy · Python
    Candidate generation

    Deterministic review prompts

  11. STAC · public catalogs
    Provider metadata

    Availability, footprints, licenses

  12. Schema 1.0.0
    Versioned analysis bundle

    Images, lineage, warnings, hashes

  13. Append-only events
    Owner assessments

    Correctable browser-local history

  14. Vite · FastAPI · Vercel
    Public EchoAtlas deployment

    Bounded prepared demonstration

surfaceservicedataaiintegrationcontrolruntime
Read system connections
  • Explore → Workflow API
  • Accessible results → Workflow API
  • Workflow API → Provider adapters
  • Provider adapters → Provider metadata
  • Explore → Pair review
  • Pair review → Immutable selection
  • Immutable selection → Pinned acquisition
  • Pinned acquisition → Local raster pipeline
  • Local raster pipeline → Candidate generation
  • Candidate generation → Versioned analysis bundle
  • Versioned analysis bundle → Analyze
  • Analyze ↔ Owner assessments
  • Workflow API → Public EchoAtlas deployment
  • Versioned analysis bundle → Public EchoAtlas deploymentapproved demo only

Selected archive / 2006–2007

yU+co studio website

An early interactive-development project built with ActionScript 2, video, XML, and multi-user technology. The studio site received a 2006 Graphic Design USA Certificate of Excellence in Communication and Graphic Design and was named a 2006 Webby Awards Honoree.

Minimal white yU+co website interface with branching navigation labeled Work, Clients, and News.
Archived interface

Carl’s contribution

Carl contributed interactive development using ActionScript 2, video, XML, and multi-user technology. The two awards recognize the site and its team rather than sole authorship by any one contributor.

More of the story

GitHub archive

Public repositories

A selection of current products, useful libraries, experiments, and earlier code. Updated . The case studies above tell the deeper product and architecture stories.

12 of 12 repositories

  • TypeScript2026

    jolene-ai

    A policy-gated personal chief-of-staff agent with Slack, governed memory, approvals, workflows, and read-only project awareness.

    • AI agents
    • Slack
    • approvals
    • memory
    Repository ↗
  • Python2026

    earth-atlas-ai

    EchoAtlas is a planned civilian SAR intelligence workbench for deterministic infrastructure-change candidates, evidence, and human-reviewed assessments.

    • FastAPI
    • React
    • TypeScript
    • SAR
    Repository ↗
  • TypeScript2026

    matchmaker-ai

    Argent is a human-led matchmaking concept spanning public and staff web apps, mobile clients, an API, background work, and framework-light domain policy.

    • Next.js
    • Flutter
    • Fastify
    • PostgreSQL
    Repository ↗
  • TypeScript2026

    jobsearch-dashboard-ai

    A local-first job-search system for role discovery, evidence management, truthful application materials, outcome tracking, and controlled workflows.

    • Next.js
    • LangGraph
    • MCP
    • PostgreSQL
    Repository ↗
  • TypeScript2026

    progression-lab-ai

    An AI-assisted music-theory product for generating chord progressions, voicings, and arrangement ideas with playback, visualization, and sharing.

    • Next.js
    • OpenAI
    • music theory
    • Tone.js
    Repository ↗Live site ↗
  • TypeScript2026

    emf-disturbance-sim

    An interactive 3D electromagnetic-field lab for simulating emitter behavior, interference patterns, and contested RF zones in real time.

    • Next.js
    • React Three Fiber
    • 3D
    • simulation
    Repository ↗Live site ↗
  • TypeScript2026

    webauthn-core

    Reusable server-side WebAuthn orchestration with pluggable adapters for challenges, credentials, and MFA state.

    • WebAuthn
    • MFA
    • authentication
    • adapters
    Repository ↗