Menu

William Etheridge

Technical founder & applied AI engineer

Download PDF Email William
+44 7931 408305wjeth@pm.meLondon, UKLinkedIn ↗

Profile

Technical founder and product-minded engineer who has built and deployed AI agents, evaluation systems, context infrastructure and customer-facing software. Works directly with startup teams to understand critical workflows and user constraints, navigate ambiguous problems, and turn customer needs and model failures into reproducible evals and durable product improvements.

Technical skills

AI systems
Agent architecture, evaluation design, benchmarking, prompt and workflow iteration, tool-use workflows, model behaviour analysis, LangGraph, LangChain, Model Context Protocol (MCP), OpenAI and Anthropic APIs
Software engineering
TypeScript, JavaScript, Next.js, Node.js, Hono, React Native, PostgreSQL/pgvector, MongoDB, Drizzle ORM, pg-boss
Infrastructure & integrations
AWS App Runner, Render, OAuth, webhooks, background workers, Notion, Google Drive, Google Calendar, Slack, Telegram/Grammy, Clerk, Stripe
Product & customer
Customer discovery, workflow mapping, rapid prototyping, technical communication, cross-functional delivery, early-stage startups, end-to-end delivery

AI & product engineering experience

AI Systems Engineer · Crispin

Independent project

2026
  • Built and privately used a Telegram-based AI assistant using a thin router and specialist agents to organise work in Notion and act through Google Calendar.
  • Designed 300+ production-faithful evals spanning routing, tool selection and arguments, full-chain behaviour and multi-turn state; used them to reproduce and isolate failure modes observed in real workflows.
  • Debugged silent partial writes, brittle calendar actions and eval-production drift; converted failures into regression coverage, then redesigned the system around production-matching schemas, batch and single-call tools, and code-enforced guarantees.
  • Benchmarked model and reasoning configurations by workflow, balancing quality, latency and cost; retained higher reasoning where cheaper settings caused agents to skip tools and answer without grounding.
Technical case study

Founder & Engineer · Latently

London, UK

Feb 2026—Present
  • Founded and launched a pilot context platform designed to reduce repeated briefing and expose source-backed project knowledge from Google Drive, Slack, Notion and uploads to MCP-compatible AI tools.
  • Chose a canonical full-text source architecture over a vector-search-first system, prioritising grounded responses, auditability and source-level access control.
  • Built OAuth ingestion, rendering and refresh pipelines across documents and channels, including Slack backfill and events, rate-limit handling, reconciliation and a separate background worker.
  • Designed the read-only MCP interface and own product discovery, architecture, implementation, testing, debugging and deployment across a TypeScript, Next.js, Hono and PostgreSQL stack.
Project detail

Customer engineering & startup delivery

Founder / Technical Product Consultant · WJETH Ventures

London, UK

May 2020—Present
  • Serve as a hands-on technical partner to startup founders and operational teams, uncovering critical workflows and constraints and translating ambiguous customer needs into technical requirements and testable success criteria.
  • Lead engagements from rapid prototype through pilot and delivery, communicating technical choices and trade-offs across technical and non-technical stakeholders.

Gather Design · Product provenance platform

  • Partnered with Gather's team to understand how provenance records fit into its production process, then translated that workflow into a full-stack system for creating digital product passports and managing furniture through production.
  • Built with Next.js, Clerk, MongoDB and Base; delivered to the customer for pilot.

MushroomRise · AI content automation

  • Translated a repetitive research-to-content process into an automated Next.js workflow that turned scientific papers into short-form content using OpenAI text and image APIs and published through the TikTok and Meta APIs.

Founder · Origin Thread

London, UK

Jan 2021—Jul 2023
  • Built a clothing product linking garments to Cardano-backed digital provenance through NFC and QR codes, including developing fabric-printed QR implementations.
  • Took the product from concept to first sales across software, supply chain, manufacturing and storytelling; travelled to India during COVID to establish suppliers, and the lessons later informed the Gather system.

Independent Mobile App Developer

Selected applications · London, UK

2023—Present
  • Built and released three React Native applications spanning habit formation, educational discovery and AI meal planning, owning product definition, implementation and App Store delivery.

Earlier experience

Digital Marketing Manager · Acrospire

Sep 2017—Apr 2020

Led a cross-functional company rebrand and the UK launch of a new solar-lighting product range.

Growth Marketing Manager · Digital Voices

Sep 2016—Aug 2017

Joined as the third team member and worked across early-stage growth and client delivery.

Delegate Sales · Worldwide Business Research

Aug 2015—Aug 2016

Became the fastest new salesperson to achieve the probation-period sales target.

Education

BMus · Royal Academy of Music

London, UK

2010—2014

Selected performance highlights include the BBC Proms at the Royal Albert Hall, European touring and recording at Abbey Road Studios.