Type: Contract
Duration: 5 months [extendable based on need and resources]
Location: Remote
Start Date: December 2025

We welcome applications from both mid-level engineers (4 to 6 years of experience) and senior engineers (7+ years of experience). Title and compensation will be commensurate with experience. Senior candidates would be expected to bring stronger architectural judgment, more independence in technical decision-making, and experience leading technical initiatives.

About Tabiya: Tabiya builds open-source software and standards to unlock economic opportunity for all. Originating from the University of Oxford, we partner with government employment services, NGOs, and job platforms to create pathways that recognize skills from all work—including informal and traditionally unseen activities. Our mission is to make labor markets more efficient, equitable, and inclusive.

Project Overview: Horizon is Tabiya's transparent, AI-powered matching and analytics engine designed to connect jobseekers with economic opportunities while providing labor market intelligence to policymakers and practitioners. We have a prototype matching algorithm developed for a research project that demonstrates core concepts of matching jobseeker skills to job opportunities using our ESCO-based inclusive skills taxonomy.

Our Taxonomy-Grounded Approach: All matching in Horizon is grounded by a version of the European Skills, Competences, Qualifications and Occupations (ESCO) taxonomy, which we've extended to include skills from informal work, household activities, and other traditionally undervalued experiences. This structured, human-vetted taxonomy serves as critical guardrails for all of Tabiya’s solutions, providing transparency and accountability. By mapping all unstructured data from both jobseekers and employers onto this standardized framework, we ensure that matches are explainable and based on validated relationships.

We are looking for a dedicated ML Engineer to bridge the gap from research prototype to a demonstrable MVP. The ML Engineer will have a clear starting point (the existing prototype) and will work closely with our existing tools including the Livelihoods Classifier, which already maps free-text job descriptions and CVs to our taxonomy.

Core Responsibilities

  1. Defining core functionalities and specifications needed to bridge prototype to MVP, in close coordination with Tabiya’s management and research teams
  2. Designing and implementing the matching algorithm (refactored from prototype or rebuilt) as clean, modular, well-documented Python code
  3. Building analytics capabilities that surface insights from aggregated data on jobseeker and opportunity side
  4. Testing with real partner data from 3-4 organizations to validate matching quality and identify improvements
  5. Building a simple demonstration interface that allows non-technical users to interact with the matching engine
  6. Deploying to a cloud platform with a straightforward, maintainable architecture (not necessarily production-grade infrastructure)
  7. Creating clear documentation for future technical teams to understand, modify, and scale Horizon as a marketable product

Technical Scope

  • Technical evaluation of existing prototype against product needs
  • Clean, modular Python codebase (production-grade, not research scripts)
  • Leverages Tabiya's existing skills taxonomy (see our documentation and Github for more details)
  • Transparent and explainable matches (users can see why matches were made)
  • Analytics layer that provides both explainable individual matches and aggregated labor market insights
  • Validated with real data from 3-4 partners with documented evaluation results
  • Simple cloud deployment accessible via URL (no local installation required)
  • Well-documented code, architecture, and evaluation methodology
  • Version controlled in GitHub (public repository)
  • Can handle 100+ jobseeker profiles and 50+ jobs for demonstration purposes

Preferred Stack

  • Backend: Python
  • Data Processing: NLP/text processing (spaCy, transformers, or similar), skikit-learn, pandas, numpy
  • Data Persistence: MongoDB (our taxonomy and Compass already rely on MongoDB Atlas)
  • Cloud Platform: Google Cloud Platform preferred.
  • Frontend/Interface: Streamlit, similar frameworks, or React.js/TypeScript/Material UI if needed. We prioritize getting to a functional demo quickly over interface polish.

Qualifications

Essential

  • 4-6 years of experience in applied ML engineering or data science
  • Strong Python development skills in writing production-grade code, not just notebooks
  • Experience evaluating and making architectural decisions about ML systems
  • NLP and text processing experience
  • Experience designing and conducting ML model evaluations (match quality, bias/fairness)
  • Cloud deployment experience (preferably GCP)
  • Practical understanding of ML explainability
  • Self-directed, makes good technical judgments with incomplete information
  • Strong documentation and communication skills

Highly Desirable

  • Experience refactoring research code or building MVPs from prototypes
  • Experience with matching, recommendation, or ranking systems
  • Familiarity with employment/labor market domains or skills taxonomies
  • Experience working with data from emerging markets or development contexts
  • Comfort working with messy, real-world data from diverse sources
  • Understanding of algorithmic bias and fairness in employment contexts

Application Process

Submit the following details to horizon-recruitment@tabiya.org by 20 November 2025. We will begin reviewing applications immediately upon receipt and may make an offer before the deadline, so early application is strongly encouraged.

  • CV/Resume
  • Cover letter (1 page max) addressing:
    • Experience evaluating and building matching systems, recommendation engines, or similar ML applications
    • Why Tabiya's mission resonates with you
  • Brief technical approach (1-2 pages): How you'd assess the existing prototype and develop it into a validated, demonstrable MVP. Include an indicative timeline (to be complete by 1 May 2026) and a budget.
  • GitHub profile or other portfolio demonstrating relevant work (especially: ML systems, deployed demos, or evaluation methodologies)
  • 2 professional references

The selection process will include a review by our team, a technical panel interview, reference checks, and a conversation with Tabiya leadership.