About Good Outcomes

Good Outcomes is building the regulatory intelligence platform for outcomes-based governance in financial services. We help financial institutions transform customer communications into actionable regulatory insights, predict outcome decisions, and proactively identify conduct risk before it becomes a problem.

Our platform processes thousands of customer signals daily, using advanced ML models to classify complaints, predict outcomes, and surface the root causes that matter most to compliance teams.

The Role

We're looking for an AI Engineer who can guide and orchestrate agents to take our classification models and ombudsman prediction systems to the next level. You'll direct work on transformer architectures, harness agents to iterate on model training, and ship ML systems that run in production at scale.

This role sits at the intersection of applied ML research and production engineering. You'll own the full lifecycle: from guiding agents through DeBERTa fine-tuning experiments to orchestrating deployment of inference services on Kubernetes. Your job is to steer, review, and refine—not to write every line yourself.

What You'll Work On

ML Models & Classification

• Own and evolve our multi-task DeBERTa v3 models for classification and predictions
• Improve our model accuracy and develop evaluation frameworks
• Build and refine training pipelines using PyTorch, HuggingFace Transformers, and custom datasets
• Develop new classification capabilities: sentiment analysis, vulnerability detection, compliance scoring

Explorer

• Enhance our current prediction model that processes large corpuses of financial decisions
• Build semantic exploration over regulatory documents using embeddings (text-embedding-3-large) and pgvector
• Develop the ML backend powering our voice agents

Production ML Systems

• Deploy and maintain FastAPI inference services for real-time classification
• Optimize model serving for latency and throughput on GKE
• Build monitoring and observability for ML systems using OpenTelemetry
• Manage model versioning, A/B testing, and gradual rollouts

You Should Have

Core ML Skills

• Deep experience with transformer architectures (BERT, DeBERTa, or similar)
• Hands-on PyTorch expertise—you've trained and fine-tuned models from scratch
• Strong understanding of NLP: tokenization, embeddings, attention mechanisms
• Experience with HuggingFace ecosystem (Transformers, Datasets, Tokenizers)

Production Experience

• Deployed ML models to production and kept them running reliably
• Familiar with model serving patterns: batching, caching, async inference
• Experience with vector databases and similarity search (pgvector, Pinecone, or similar)
• Comfortable with Python backend development (FastAPI preferred)

Agentic Engineering (This is non-negotiable)

• You work daily with Claude Code, Codex, or whichever agent fits the task—and you know when to reach for each
• You have a feel for which agent harness works best for which problem: experiment design, code generation, debugging, documentation
• You think of yourself as a guide—directing agents through ML experiments and reviewing their output
• You bring your own agentic skills to the table as part of your contribution to the team
• You continuously evolve your workflow as new capabilities emerge
• You've built with LangChain, LangGraph, or similar LLM orchestration frameworks

Bonus Points

• Experience in financial services, regtech, or compliance domains
• Familiarity with UK regulatory frameworks (FCA, FOS, Consumer Duty)
• Background in multi-task learning or transfer learning
• Experience with voice AI systems (LiveKit, WebRTC)
• Contributions to open-source ML projects

Tech Stack

ML/AI: PyTorch, HuggingFace Transformers, DeBERTa v3, scikit-learn, OpenAI API, LangChain, LangGraph Backend: Python 3.11+, FastAPI, Celery, PostgreSQL with pgvector, Redis Infrastructure: Kubernetes (GKE), Docker, Terraform, Cloud Build Observability: OpenTelemetry, LangSmith, Google Cloud Monitoring

Why Good Outcomes?

• Impact: Your models directly help financial institutions treat customers fairly and avoid regulatory harm
• Technical depth: Work on genuinely hard ML problems with real-world constraints
• AI-native culture: We build with AI tools, not just build AI tools
• Small team, big ownership: You'll own entire systems, not just tickets
• Remote-first: Work from anywhere, async-friendly culture

How to Apply

Install our application skill, then ask your agent to apply for the AI Engineer - ML & Classification Systems role.

Option 1: Via skills.sh (Recommended)

The fastest way to install:

npx skills add DataEQ/go-people-skills

This works with Claude, Codex, Cursor, Moltbot, and any other harness that supports skills.

Option 2: Claude Code Plugin Marketplace

Add the marketplace and install the plugin:

/plugin marketplace add DataEQ/go-people-skills
/plugin install apply-to-good-outcomes@go-people

Option 3: Direct Clone

git clone https://github.com/DataEQ/go-people-skills.git ~/.claude/plugins/go-people-skills

Usage

Once installed, just ask your agent to apply for this role—or run /apply to get started.

Good Outcomes is committed to building a diverse team. We encourage applications from candidates of all backgrounds.