Claims linked to evidence

Capabilities → evidence

Browse the capabilities first, then inspect the evidence behind each one. Every reference opens a full artifact record without taking you away from the list, and the complete index remains below. No artifact, no claim.

01 / References

Browse by capability

Select a reference to inspect it without leaving the list.

Open a group, then a capability. References are grouped by evidence type, and each E-number points to one canonical record in the index.

Core The capabilities most of my work depends on. 11 capabilities · 63 references
Shipped production agentic/LLM systems 12 artifacts
Case studies
E19E20E22
Live demos
E01E02E03
RAG end-to-end 6 artifacts
Case studies
E19E21
Live demos
E01E02
Code
E09E26
Eval harnesses 7 artifacts
Case studies
E20E21
Live demos
E01E02
Code
E11E17
Evaluations
E12
Observability 2 artifacts
Case studies
E22
Code
E07
Cost/performance engineering 6 artifacts
Case studies
E19
Live demos
E03
Code
E05E07
Evaluations
E12
Briefs and diagrams
E13
Structured outputs + schema validation 7 artifacts
Docker + CI/CD + git hygiene 3 artifacts
Production Python 5 artifacts
Business translation 4 artifacts
Case studies
E19
Code
E17E18
Briefs and diagrams
E25
End-to-end lifecycle ownership 4 artifacts
Case studies
E19E20E21
Code
E17
Reliability, retries & graceful degradation 7 artifacts
Frequently required Capabilities that recur across briefs and roles. 11 capabilities · 45 references
LangChain / LangGraph / named frameworks 3 artifacts
Code
E06E08
Briefs and diagrams
E23
Cloud AI platforms 4 artifacts
Code
E06E14
Briefs and diagrams
E13E24
Workflow automation tools 3 artifacts
Case studies
E22
Code
E33
Briefs and diagrams
E25
API / MCP / integration breadth 8 artifacts
Case studies
E22
Briefs and diagrams
E25
Postgres / pgvector 3 artifacts
Case studies
E19
Code
E09E26
Human-in-the-loop + guardrails 8 artifacts
Case studies
E19E20
Live demos
E03
Briefs and diagrams
E25
Data governance / sovereignty 4 artifacts
Case studies
E19
Conversational AI 6 artifacts
Case studies
E20E21
Live demos
E01E02E03
Code
E30
Discovery / pilot / POC craft 2 artifacts
Case studies
E19E21
Terraform / IaC 2 artifacts
Code
E14
Briefs and diagrams
E24
Classical ML / statistical modelling 2 artifacts
Code
E27E28
Context The tooling practice the rest of this work is built with. 1 capability · 1 reference
AI-assisted engineering (Claude Code) 1 artifact
Case studies
E22
02 / Artifacts

Artifact index

One canonical record per artifact.

The 33 artifacts behind the register, each with its kind, the claim it supports and a short summary. Each title links to the artifact itself.

E01live demo

Live grounded RAG chatbot: education prospect demo A

Deployed a grounded RAG chatbot that passed a 33/33 golden-QA gate before ship.

Website rebuilt + chatbot answering only from the business's real content; eval suite re-run against the live URL post-deploy.

E03live demo

Agora: feedback-analysis product demo

Built a public demo where anyone can watch a real traced pipeline run step through its stages, then run the actual dashboard themselves via a one-command Docker bundle.

Live product case-study site with an animated real-trace pipeline stepper and a runnable Docker demo.

E04repo

agora-demo: runnable Docker bundle

Clone → docker compose up → working analysis dashboard in ~5 minutes.

Public compose bundle; pipeline IP excluded via verified sanitized images.

E06code exhibit

AWS Bedrock + LangChain structured outputs

Integrated Bedrock via langchain_aws so structured outputs pass schema validation on arrival, and built a throttling ladder tuned to Bedrock's actual per-minute quota instead of a generic backoff curve.

Three fast retries then a deliberate 61s wait past the quota boundary, jittered; typed and string-wrapped ThrottlingException both caught; timeouts handled on a separate shorter path. Part of an AWS migration incl. Bedrock evaluation and GPU self-hosting of Llama 70B.

E07code exhibit

Multi-provider model router + token/cost ledger

Requests fall back through an ordered provider chain on failure, retry against schema validation, and every call's cost is logged automatically.

Durable retry-payload queue on exhaustion; free-model-only governance for autonomous surfaces.

E08code exhibit

Workflow dispatcher with human gates

Workflows pause mid-stage for human approval and resume from the gate.

LangGraph-equivalent constructs implemented from scratch (see equivalence diagram).

E09code exhibit

Hybrid RAG retrieval (RRF) on Cloudflare Workers

Keyword and vector retrieval fused by RRF, degrading gracefully rather than failing, and running in the live demos today.

Sibling server-side implementation uses Postgres FTS + pgvector hybrid search.

E10code exhibit

MCP server (FastMCP, Streamable HTTP)

Built and shipped a hand-written MCP server (5 tools, 3 resources, 3 prompts) with auth and per-tool rate limits enforced.

From a consumer AI product in private beta (capabilities disclosed, concept not).

E11code exhibit

Golden-QA eval loop (evals as ship gates)

Chatbots must pass the full golden suite before and after deploy.

must/mustNot assertions incl. leak guards; sample fictional-client case set included.

E12eval report

LLM-judge eval report + cost tracker

Picked the shipped model against explicit pass/fail gates on quality, cost, and latency, with the report to show the reasoning.

Golden corpus, 5-dimension rubric, p50/p95 tracking, CSV export.

E13brief

Model pricing analysis (cost engineering)

Derived per-job cost from token mechanics; caught a 6-7x pricing error in code constants.

Standard vs batch/flex tables across six models, with reconciliation addendum.

E14code exhibit

Terraform (GCP Cloud Run, Pub/Sub, IAM)

Authored IaC for a production-shaped product; separately extended an upstream AWS stack and used selected AWS services for development and performance work.

State/vars excluded; AWS (ECS/Batch/SQS) stack credited as upstream, extended by me.

E16code exhibit

Docker image leak scanner

Proves proprietary code is absent from every published image layer.

docker save + per-blob marker scan; avoids the rm-in-a-later-layer trap.

E17repo

Document generation pipeline (skills-as-code)

Generates tailored documents that never invent a fact, with keyword-coverage checks and a human checkpoint before anything ships.

Never-invent rule, two-skin design system, ATS-safe rendering, full test suite.

E19case study

Case study: Stoa: sovereign consultation analysis

Built the consultation-analysis application solo: pipeline, backend, review frontend, infrastructure and docs; delivered it for a paying client, then took it into a 2026 government consultation of more than 2,000 comments on self-hosted open-weight models inside New Zealand.

Configurable multi-stage pipeline, hierarchical themes, audit-trailed human review, NZISM-aware RFP work, and self-hosted open-weight inference run on New Zealand research compute.

E20case study

Case study: Agora: feedback analysis product

Productised the platform pattern; on a public academic benchmark (SemEval-2014), scored in-house, topic detection comes to 0.83 F1 in the product's discovery mode. The client-engagement validation is the stronger evidence, and the case study ranks them.

Discovery-mode detection 0.83 F1 (product-as-sold, no gold categories seeded); predetermined-mode polarity roughly level with the best 2014 system (0.84 vs 0.829, like-for-like); demo ships separately.

E21case study

Case study: website→demo delivery pipeline

Turns a prospect's URL into a live grounded chatbot and a priced proposal, through 7 human-gated stages.

Intake → scrape → design → build → chatbot evals → ship → proposal; two live instances.

E22case study

Case study: FridaiOS: an operating layer for AI-native work

Built a provider-agnostic, file-system-first agent platform with the same shape as an organisational agent deployment, and run my own daily work through it.

Skills registry, staged workflows with human gates, retrieval with golden tests, free-tier model routing with an append-only cost ledger.

E23diagram

FridaiOS ↔ LangGraph equivalence map

Mapped my orchestration engine to LangGraph primitives, concept for concept, so the equivalence can be checked directly.

nodes/state/checkpoints/tools ↔ stages/run-state/pause-gates/skills.

E24brief

Cloud portability brief (AWS ↔ Azure)

Mapped AWS services used in development work, plus upstream AWS foundations extended, to Azure equivalents with provenance stated per service.

Bedrock↔Azure OpenAI/Foundry and EC2↔NC-series derive from development/performance work; five further mappings are upstream foundations extended; Azure remains architecture-only.

E25diagram

Admissions operations automation

Built the n8n automation layer joining enquiry, CRM, calls, follow-up, enrolment records and Student Services handover across a distributed admissions operation.

Selective AI for language and transcript work inside a rules-driven customer journey; enquiry replies moved from 24+ hours to about 10 and enrolment handover from two working days to the same day; the nine workflows are rebuilt on synthetic data as Exhibit 20.

E26code exhibit

Exhibit 13: legislation RAG (pgvector + citations)

Extracted and sanitized the legislation RAG built for Stoa: section-level bill corpus, hybrid retrieval, citations a reviewer can verify.

Extracted from the consultation-analysis application: legislation parsed into a section-level corpus, hybrid retrieval over it, and theme-to-section mappings grounded in verifiable citations.

E27code exhibit

Exhibit 14: conversation feature engineering + cluster search

Segmented ~800 recorded sales calls on conversational shape and content, then tested the segments against real conversion outcomes.

TF-IDF reduced before it meets hand-built conversation features; silhouette search over feature subsets and k. Ships with a synthetic fixture; the real transcript corpus is customer PII and is excluded.

E28code exhibit

Exhibit 15: model selection under a weak signal

Found that tuned ensembles beat a linear baseline by nothing, and shipped the linear model with learning curves as the evidence.

Linear vs tuned Random Forest vs XGBoost in one pipeline, learning-curve overfitting diagnosis, and a one-standard-error rule that declines to prefer a lead smaller than its own measurement noise.

E29code exhibit

Exhibit 16: graceful degradation under failure

When the model can't place a response, the pipeline degrades how specific the answer is rather than inventing one: a fallback ladder that climbs the theme hierarchy and then stops, recording the response as unmapped instead of fabricating a finding.

From Stoa's unattended analysis runs over tens of thousands of public submissions. Leaf mapping, then a lenient re-prompt, then progressive fallback up the parent hierarchy, then a deliberate refusal to force a mapping. Ships alongside the provider-level retryability classifier that separates transient failures from terminal ones.

E30code exhibit

Exhibit 17: realtime voice agent

Built the portfolio voice agent first on OpenAI Realtime/WebRTC, then migrated the live implementation to Gemini Live/WebSockets, retaining streamed audio, interruption handling, transcription and short-lived session credentials.

Versioned transport evidence: earlier OpenAI WebRTC implementation; current Gemini Live WebSocket implementation; provider keys remain server-issued and short-lived.

E31code exhibit

Exhibit 18: consent-gated agent actions

Built an agent-action boundary where models may draft but cannot execute: actions require visible user confirmation, signed authorization, strict validation and idempotent server-side recording.

Draft-only tool contracts, a separate confirmed-action envelope, short-lived signed authorization, durable two-action session limits and failure handling that records notification status without duplicating the external effect.

E32code exhibit

Exhibit 19: private context isolation

Built isolated per-recipient AI context using hashed bearer exchange, signed HTTP-only sessions, allowlisted browser projections, integrity validation, expiry and revocation.

The raw bearer is never stored; only its SHA-256 hash reaches D1. Every session restore revalidates packet integrity, expiry and revocation, while private notes and source material remain server-side.

E33code exhibit

Exhibit 20: n8n admissions automation (rebuilt)

Rebuilt the nine admissions workflows as n8n files that import and run without credentials: every service and model call carries pinned synthetic output, and every rule runs for real.

Rule nodes check model output against the course sheet before a person sees it, human gates hold the run until an adviser or Admissions responds, and one script imports all nine into a fresh n8n, runs every branch and compares each rule's output with its expected file.