About Lighthouse Canton
Founded in 2014, Lighthouse Canton is a Singapore-headquartered global investment management firm with approximately US$ 5+ billion in Assets Under Management (AUM). Lighthouse Canton is one of the leading and award-winning investment firms in the region, focusing on managing funds with uncorrelated investment strategies, providing investors with the opportunity to achieve consistent returns across market cycles. Lighthouse Canton has offices in Singapore, Dubai, India, and the UK.
Job Purpose
We are offering an AI Engineering Internship designed as a structured, mentored introduction to building production-grade AI systems inside a regulated wealth management firm. You will sit inside our small AI Lab and work alongside experienced engineers on real internal applications across Investment, Relationship Management, Risk, Operations, and Compliance - not toy projects, and not coffee runs.
What You Will Do
Building & Supporting AI Applications
- Pair with senior engineers on AI-powered internal applications using LLMs, RAG pipelines, and agents. Expect to start by reading code and small fixes, and ramp up to owning well-scoped features under guidance.
- Help build and tune retrieval systems: chunking strategies, embeddings, vector search, and basic re-ranking.
- Practise prompt engineering and context engineering—structuring inputs, managing memory, and compacting context - to keep agents coherent across multi-step tasks.
- Help wire internal knowledge bases, market data feeds, and operational systems into agents via Model Context Protocol (MCP) servers and other typed-tool interfaces.
Agent Harness & Evaluation
- Get hands-on with the agent harness - the runtime layer around the LLM that manages execution loops, tool calls, state, memory, and error handling. You will be expected to learn the vocabulary (ReAct, planner/executor, guardrails, permission gates, loop budgets) and apply it in real code.
- Help build evaluation harnesses: writing test cases, regression suites, deterministic checks (schema validation, citation checks),and simple LLM-as-judge graders—so we can tell whether changes actually improve quality.
- Instrument agents with basic observability: tracing tool calls, logging token usage, latency, and outcomes; help triage failure traces and feed lessons back into prompts and tool specs.
- Contribute to internal documentation on harness patterns, evals, and lessons learned - useful both to the team and to your own learning.
Collaboration & Knowledge Sharing
- Partner with stakeholders in Investment, Relationship Management, Risk, Operations, and Compliance to understand requirements—always alongside a senior team member.
- Take part in code reviews, design discussions, and weekly team check-ins. Asking questions is part of the job, not a sign of weakness.
- Present a short end-of-internship readout summarising what you built, what you learned, and what you would do differently.
What We Are Looking For
Technical Foundations
- Programming: Comfortable in Python from coursework, personal projects, or prior internships. Any exposure to JavaScript/TypeScript, Git, and the command line is a plus.
- AI/ML Curiosity: Has built at least one small project involving LLMs (e.g., via OpenAI, Anthropic, or open-source models) - a chatbot, a RAG demo, a hackathon project, a tinkering side project, an agent that does something silly. Quality of curiosity matters more than scale.
- Agentic AI Awareness: Familiar at a conceptual level with agent frameworks (e.g., LangChain, LangGraph, CrewAI, OpenAI Agents SDK, Claude Agent SDK) and modern terms like MCP, RAG, evals, and guardrails. We do not expect mastery - we expect that you have read about them and want to go deeper.
- AI-Assisted Development: Comfortable using coding assistants such as Claude Code, Codex, or GitHub Copilot as part of how you build - and aware that they need supervision, tests, and clear specs to produce reliable code.
- Data & Web Basics: Light exposure to databases (SQL, or any NoSQL store), and to web/API basics (HTTP, JSON, a back-end framework such as FastAPI or Flask) is helpful but not required.
Mindset & Soft Skills
- Genuine curiosity about AI, finance, and the messy intersection of the two. You should be excited to learn how a wealth management firm actually works.
- Comfort with ambiguity: real projects rarely arrive with neat specs. We will help you scope them, but we want someone willing to sit with uncertainty.
- Communication: able to explain technical ideas in plain English (and in writing), and to flag blockers early rather than getting stuck quietly.
- Integrity: this is a regulated environment. We need an intern who treats client data, MNPI, and internal information with the same care a full-time hire would.
Education
- Currently pursuing a Bachelor’s or Master’s degree in Computer Science, Artificial Intelligence, Data Science, Software Engineering, Quantitative Finance, or a related field—or a recent graduate within 12 months.
- No prior professional experience required. Strong coursework, personal projects, open-source contributions, hackathon work, or research projects are all valid signals.
- Penultimate-year and final-year students are encouraged to apply.
