New opportunity

Research Assistant, BAC Agentic AI and Decentralized Finance

National University of Singapore · Singapore

About this role

The NUS Business Analytics Centre seeks a Research Assistant for a project developing autonomous AI agents that can reason, plan, interact with blockchain ecosystems and execute financial workflows securely and efficiently. Based at Kent Ridge Campus, the researcher will work with faculty, industry collaborators and graduate researchers on AI agent architectures for Web3 applications.

Responsibilities

  • Design and implement AI agent frameworks and autonomous or multi-agent systems for blockchain-based tasks.
  • Integrate agents with Ethereum-compatible infrastructure, and develop modules for smart contracts, wallets and DeFi protocols. Implement retrieval-augmented generation, memory systems, tool calling and agent orchestration.
  • Review research on AI agents, LLMs, blockchain and DeFi. Benchmark and experiment with agents, including security testing.
  • Translate industry requirements into research prototypes; prepare technical reports, publications and documentation; and support demonstrations and technology transfer.

Qualifications

  • A bachelor’s degree in a relevant field, experience with modern AI development frameworks, and good knowledge of blockchain, smart contracts and DeFi.
  • Proven experience delivering industry-facing or commercial AI solutions, plus strong communication skills and the ability to work with non-technical stakeholders. No minimum number of years is specified.

Beneficial experience includes multi-agent orchestration, planning and reasoning, workflow automation, RAG or Model Context Protocol; Ethereum, Solidity, Hardhat or Foundry, Web3.py, Ethers.js and wallet integration; or DeFi concepts such as automated market makers, decentralized exchanges, lending, staking, cross-chain protocols, liquidity pools and tokenomics.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

AI agentsMulti-agent systemsAgent orchestrationTool callingFunction callingWorkflow automationPlanning algorithmsReasoning algorithmsRetrieval-Augmented GenerationModel Context ProtocolLarge Language ModelsBlockchainWeb3EthereumSolidityHardhatFoundryWeb3.pyEthers.jsSmart contractsWallet integrationDecentralized FinanceAutomated Market MakersDecentralized ExchangesLending protocolsYield farmingStakingCross-chain protocolsLiquidity poolsTokenomics

Your skill match

Checking your profile…

YOUR NEXT STEP

Get interview-ready for this role

A focused preparation guide, built around this job’s responsibilities and requirements.

✦ AI-generated guide
Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.
01

Understand the role

What matters most

  • Build agents that can act

    The role involves designing agent frameworks and autonomous or multi-agent systems that plan, reason and execute blockchain-based tasks.

  • Connect agents to Web3 systems

    A central development duty is integrating agents with Ethereum-compatible infrastructure and building modules for smart contracts, wallets and DeFi protocols.

  • Implement agent capabilities

    The posting specifically calls for retrieval-augmented generation, memory systems, tool calling and agent orchestration within the research systems.

  • Evaluate performance and security

    The researcher will benchmark and experiment with agents, conduct security testing and review relevant AI, blockchain and DeFi research.

  • Turn research into usable outputs

    The role includes translating industry requirements into prototypes with collaborators, then producing reports, publications and documentation and supporting demonstrations.

Likely challenges to think through

  • Autonomy versus safe execution

    Preparation hypothesis: Giving an agent freedom to execute financial workflows may conflict with the project's security aims. Prepare to explain how you would bound tool access and test contract or wallet interactions before allowing execution.

  • Agent sophistication versus measurable gains

    Preparation hypothesis: Multi-agent orchestration and memory can add complexity without improving outcomes. Prepare a comparison plan with simpler baselines, task-level measures and repeatable experiments.

  • Industry utility versus research rigor

    Preparation hypothesis: A prototype must address collaborator requirements while yielding defensible research findings. Prepare to describe how you would scope a demonstrable workflow, document assumptions and report evaluation results.

03

Practice your answers

Technical practice

How would you design an agent to plan and execute a blockchain-based financial workflow?

What to coverThe role involves agents that reason, plan and execute blockchain-based tasks. Suggested preparation: describe how you would separate planning from transaction execution, verify intermediate results and decide when an action needs approval.

When would a multi-agent design be preferable to a single agent?

What to coverThe project includes autonomous and multi-agent systems. Suggested preparation: compare task division, coordination overhead and failure handling for a Web3 workflow, explaining what evidence would justify the added complexity.

How would you connect an agent to an Ethereum-compatible network, wallet and smart contract?

What to coverThe role calls for integrations with Ethereum-compatible infrastructure and modules for wallets and smart contracts. Suggested preparation: walk through reading contract state, constructing and submitting a transaction, and checking its outcome; explain how you would handle network or transaction errors.

What would an agent need to understand before using a DeFi protocol?

What to coverThe posting names DeFi protocol interaction and lists AMMs, DEXs and lending among beneficial areas of experience. Suggested preparation: choose one protocol type and explain its core mechanics, the information needed before an action, and the trade-offs or risks the agent should surface.

How would you use retrieval and memory in a Web3 agent?

What to coverImplementing RAG and memory systems is a stated duty. Suggested preparation: distinguish retrieved reference material from persistent workflow state, then explain how you would check relevance, freshness and provenance before the agent acts on either.

How would you control tool calling in an agent workflow?

What to coverThe role includes tool calling and agent orchestration; function calling and MCP appear as beneficial experience. Suggested preparation: explain tool selection, input validation and handling failed calls, making clear which safeguards you propose rather than treating them as specified project requirements.

How would you security-test an agent that can initiate DeFi actions?

What to coverSecurity testing and secure financial-workflow execution are explicit parts of the project. Suggested preparation: identify plausible failures such as manipulated tool inputs, unintended transactions or excessive wallet permissions, and explain how tests would detect and limit them.

How would you benchmark competing agent designs for blockchain tasks?

What to coverThe researcher will evaluate agents through benchmarking and experimentation. Suggested preparation: define representative tasks and measures for correctness, efficiency and security, then describe how you would compare designs under consistent conditions and inspect failures.

How would you turn a literature review into a research experiment?

What to coverThe role includes reviewing AI-agent, LLM, blockchain and DeFi research. Suggested preparation: explain how you would identify a testable gap, establish a comparison approach and document limitations without claiming a result in advance.

Behavioural practice

Tell me about a time you translated a stakeholder's needs into a technical prototype.

What to coverThe role involves translating industry requirements into research prototypes and communicating with non-technical stakeholders. Suggested preparation: use a real situation, your actions to clarify requirements and the resulting prototype or decision; distinguish stakeholder feedback from your own assumptions.

Describe a time you worked across research and engineering perspectives.

What to coverThe researcher will work with faculty, industry collaborators and graduate researchers. Suggested preparation: give a real example of differing priorities, how you coordinated decisions and what resulted; do not imply experience with these particular collaborators.

Tell me about a time you delivered an AI solution for an external user or organisation.

What to coverThe posting asks for experience delivering industry-facing or commercial AI solutions. Suggested preparation: explain the actual user need, your contribution, how you assessed whether the solution worked and the outcome, without inventing metrics.

Describe a time an experiment or technical evaluation challenged your initial approach.

What to coverBenchmarking, experimentation and security testing are stated duties. Suggested preparation: present a real situation, the evidence you examined, the change you made and the result, including any unresolved limitation.

Tell me about a time you explained a technical result to a non-technical audience.

What to coverThe role includes reports, documentation, demonstrations and communication with non-technical stakeholders. Suggested preparation: describe the audience, how you adapted the explanation, what you did to check understanding and what happened afterward.

02

Sharpen your knowledge

Topics to prepare

  • Agent planning and orchestration

    Sketch how an autonomous or multi-agent system would plan, reason and coordinate a blockchain task; identify where a human should review an action. The posting explicitly calls for agent frameworks and blockchain-based planning and reasoning.

  • RAG, memory and tool calling

    Compare what an agent should retrieve, retain in memory or obtain through a tool call during a DeFi workflow. These components are explicitly listed among the implementation duties.

  • Ethereum, smart contracts and wallets

    Trace a proposed transaction from a wallet through a contract call to its on-chain result, noting where authorization or transaction errors could arise. The role requires blockchain and smart-contract knowledge and calls for these integrations.

  • DeFi workflow mechanics

    Work through a hypothetical swap using an automated market maker and liquidity pool, explaining the inputs, outputs and risks an agent would need to understand. DeFi knowledge is required; AMMs and liquidity pools are listed as beneficial areas.

  • Agent evaluation and security

    Define measures for task completion, efficiency and unsafe actions in a blockchain-agent experiment. Benchmarking, experimentation and security testing are explicit duties; the particular measures are preparation suggestions.

Skills to revise

  • Build a bounded agent prototype

    Implement a small agent that plans a blockchain-related task, calls a tool and records its steps without executing live financial transactions. This rehearses the stated agent-framework and orchestration duties; the bounded exercise is a preparation suggestion.

  • Practice contract and wallet integration

    On a local or test environment, read a contract value and prepare a wallet-mediated transaction; inspect errors and avoid using real funds. The posting calls for contract and wallet modules; Solidity, Hardhat or Foundry, Web3.py and Ethers.js are beneficial examples, not mandatory tools.

  • Implement retrieval and memory

    Create a small RAG workflow over technical reference notes, add a simple memory record, and check whether retrieved information supports the agent's answer. RAG and memory implementation are explicit duties.

  • Run a reproducible evaluation

    Write a few representative task cases, record outcomes and failure modes, and include an adversarial or unauthorized-action case. This practices the posting's benchmarking, experimentation and security-testing duties.

  • Explain a research prototype

    Draft a one-page technical report that turns a hypothetical industry requirement into a prototype scope, evaluation result and plain-language demonstration. This reflects the stated collaboration, documentation and non-technical communication responsibilities.

04

Your preparation checklist

  • Prepare a relevant project walkthrough

    If available, choose an AI solution or agent project and prepare a five-minute walkthrough of its architecture, your contribution, results, and how requirements were translated into a working system. The posting specifically seeks experience delivering industry-facing or commercial AI solutions.

  • Sketch an agent-to-blockchain design

    For a hypothetical DeFi task, draw an agent workflow showing planning, tool calls, wallet and smart-contract interactions, and human approval before consequential actions. Be ready to explain which parts are proposed design choices rather than past work.

  • Demonstrate a focused technical example

    Prepare code or a diagram from existing work, if available, showing RAG, memory, agent orchestration, or Ethereum-compatible integration. Explain key decisions and limitations; these are development areas named in the posting.

  • Define an evaluation and security plan

    Draft a one-page plan with task-success measures, baseline comparisons, failure cases, and tests for unsafe tool use or unintended transactions. The role includes benchmarking, experimentation, and security testing.

  • Summarize relevant research

    Select two or three papers or technical approaches on AI agents, LLMs, blockchain, or DeFi. Prepare a brief comparison of their methods, limitations, and a question worth testing, reflecting the role’s literature-review duties.

  • Gather evidence for the stated qualifications

    Organize concise examples of work with modern AI frameworks, blockchain, smart contracts, DeFi, and non-technical stakeholders, plus relevant degree details. Distinguish hands-on experience from conceptual knowledge.

  • Draft questions for the research team

    Ask which Web3 workflows the project prioritizes, how prototypes will be evaluated, and what technical reports, demonstrations, or industry-facing outputs are expected. These questions follow the advertised research and collaboration duties.

AI-generated from the job information · Prepared Sep 30, 2026 (UTC)Useful preparation, not a guarantee of eligibility or an offer.

Availability last checked: 4 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs
Finding similar opportunities…