Adjetei Martey

Open to AI / ML engineering roles · remote

AI engineer. Ships things people use.

Software engineer with an MSc in Artificial Intelligence (Distinction). Since 2019 I've shipped production software, from recommendation models on live customer data to LLM apps with retrieval, tool-calling agents and structured extraction.

Experience

  1. Apr 2026 – now

    CTO · Rumate

    • Own data, scoring and ML at a startup that rates Toronto rental buildings
    • Built the Toronto Open Data pipeline that keeps 3,000+ buildings and their inspection history up to date
    • Built the People's Score: five public sources per building, with Claude and Qwen classifying reviews to discount solicited praise. Its first run scored 100 buildings across a city ward
  2. 2022 – 2026

    Senior Software Engineer · Nguvu Global

    • Led the design and deployment of recommendation models that target a base of around 10,000 customers from their transaction data
    • Built the product's first generative AI feature: a customer support chatbot prototype on Anthropic's API
    • Cut database query times by about 40% by indexing high-traffic queries
  3. 2019 – 2022

    Software Engineer · Nguvu Global

    • Designed and shipped the first recommendation model for customer targeting
    • Built customer-facing platform features in a six-person engineering team

What I've built

  • Unstuck An AI to-do app for people who freeze at a long list.
  • FrontDesk Booking agent. Answers your guests' questions and takes bookings day and night, and hands the tricky ones to you.
  • TriagePilot Tool-calling agent. Sorts your support inbox and drafts replies for your team to approve.
  • docuchat RAG. Lets your staff get answers from your own documents in seconds, with sources.
  • Extractly Structured extraction. Turns piles of invoices and forms into spreadsheet-ready data, no retyping.

Research & ML

  • Chest X-ray classifier · Computer vision

    DenseNet121 (the CheXNet architecture) fine-tuned on NIH ChestX-ray14 to spot 14 findings at once, trained with BCE loss and evaluated with per-label AUROC.

    • PyTorch
    • DenseNet
    • Multilabel
    GitHub
  • Facial emotion recognition · Computer vision

    Reads seven emotions from face images in FER2013. A VGG19 fine-tuned with class-weighted loss reached 68.6% accuracy, against 52.7% for a custom CNN built from scratch.

    • PyTorch
    • fastai
    • Transfer learning
    GitHub

Hiring an AI engineer?
Let's talk.

Send me the role and I'll come back to you quickly. My CV has the full detail.

hello@adjetei.com