全职 AI Research - Development Intern 工作, 薪水, Simplism.io 公司招聘中 - Ricebowl

AI Research & Development Intern

Company: *************

Location: Singapore

Working arrangement: Hybrid

Employment type: Internship

Duration: 3-6

Hours: Negotiable

Compensation: Paid (S$1200)

Start date: asap

About *************

************* develops AI-supported communication technology for sales, marketing and customer engagement teams. Organizations communicate with leads and customers through channels such as email, LinkedIn, phone, WhatsApp, online advertising, events, website forms, chatbots and partner referrals. Selecting the right workflow for each interaction can be difficult because performance depends on factors including industry, company size, job role, lead source, previous conversations, intent signals, timing, channel cost and sales capacity. Our AI agent, *************, analyses communication data and supports teams with tasks such as spam detection, lead identification, sentiment analysis, conversation summarisation and response recommendations. We have already developed an initial platform and are looking for an intern to help research, test and improve the AI models behind it.

The internship

As an AI Research & Development Intern, you will work on an applied research project involving message analysis, AI safety and intelligent communication support. You will evaluate existing models, prepare datasets, design experiments and build a proof of concept that demonstrates how AI can transform lead and conversation data into practical recommendations.

The internship combines research, software development and model evaluation. You will work with guidance from the ************* team and document your findings as part of an academic or professional internship assignment.Responsibilities

You will:

• research and compare AI models for communication-analysis use cases;
• prepare, clean, label or simulate suitable datasets;
• design evaluation criteria and testing methods;
• train, configure or fine-tune selected models where appropriate;
• evaluate model accuracy, speed, privacy, cost and reliability;
• experiment with classification models, large language models, embeddings, retrieval techniques and guardrails;
• design a proof-of-concept architecture;
• develop a simple demonstration application;
• document experiments, limitations and recommendations;
• present the final results to the ************* team.
  

Research areas

Depending on the agreed project scope, your work may cover one or more of the following areas:

Content filtering

Detecting inappropriate, harmful or prohibited content in business communication.

Prompt-injection detection

Identifying messages intended to manipulate, bypass or compromise an AI agent.

Spam detection

Distinguishing unwanted, irrelevant or automated messages from legitimate communication.

Phishing detection

Recognising suspicious messages, links and social-engineering attempts.

Fraud detection

Identifying communication patterns that may indicate scams, impersonation or fraudulent behaviour.

Lead detection

Recognising messages that indicate commercial interest or a potential sales opportunity.

Sentiment and conversation-state detection

Identifying tone, urgency, satisfaction and other relevant conversational signals.

Rolling conversation summaries

Maintaining concise summaries of long conversations while retaining important context, decisions and follow-up actions.

Message suggestions

Generating relevant, accurate and context-aware response recommendations for users.

Proof-of-concept scope

The proof of concept should demonstrate how lead and communication data can be converted into recommended actions or workflows.

The project may include:

• a data model for leads, accounts, conversations, touchpoints, workflows and outcomes;
• datasets based on metadata and metrics available within the ************* platform;
• a lead-identification or lead-scoring component;
• a channel or workflow-ranking component;
• a next-best-action recommendation concept;
• a locally hosted or privacy-conscious language-model component;
• a simple interface in which a user enters lead or conversation information and receives a recommendation.

The following are outside the internship scope:

• production-ready CRM integration;
• automated outreach to real prospects;
• autonomous sending of messages;
• full production deployment;
• production-grade communication-channel automation.
  

Expected deliverables

The final deliverables are expected to include:

• a research report;
• a working proof of concept;
• model and architecture documentation;
• evaluation results;
• a description of limitations and risks;
• recommendations for further development;
• a final presentation or demonstration.

The research report should explain the problem definition, dataset, selected models, architecture, experiments, evaluation approach, results and conclusions.

Candidate profile

This internship may suit you when you are studying in a relevant field such as:

• artificial intelligence;
• computer science;
• data science;
• software engineering;
• machine learning;
• another related technical programme.

Useful experience includes:

• Python;
• machine learning;
• natural language processing;
• large language models;
• data preparation and evaluation;
• APIs or backend development;
• Git and collaborative software development.

Experience in every area is not required. We are mainly looking for someone who is analytical, curious, able to work methodically and interested in testing AI on real business problems.

What we offer

• an applied AI research assignment with a working product;
• access to relevant platform knowledge, technical guidance and feedback;
• the opportunity to influence the future development of *************;
• experience with real-world AI evaluation, safety and communication use cases;
• [internship compensation];
• [equipment, travel allowance or other benefits, where applicable].
  

Application

Please submit:

• your CV;
• a short explanation of your interest in the internship;
• your availability and preferred start date;
• relevant academic projects, GitHub repositories or portfolio examples, where available.

Applications are assessed based on relevant skills, motivation and suitability for the project. ************* considers qualified applicants without discrimination based on characteristics protected by applicable law.