12 AI-Native Engineering Project Ideas for B.E. and B.Tech Students
AIYING | AI-Native Engineering Project Lab
From AI applications to AI engineering.
Artificial intelligence is changing the nature of engineering. The opportunity for engineering students is no longer limited to building prediction models, classification systems or conventional chatbots.
Today's AI technologies make it possible to develop systems that use tools, retrieve knowledge, plan tasks, generate code, evaluate results, collaborate with humans and perform increasingly complex workflows.
This creates an exciting opportunity for engineering education: students can learn AI engineering by building AI-native systems.
Introducing the AIYING AI-Native Engineering Project Lab
AIYING proposes a collection of industry-oriented AI engineering projects for Indian B.E. and B.Tech students.
The projects are intended for teams of three to four students, working over two academic semesters as part of their major or final-year engineering project.
The objective is not simply to demonstrate an AI model or connect an application to an AI API. Students should gain experience designing, developing, integrating, testing and evaluating intelligent software systems.
Each project presents a meaningful engineering challenge, with opportunities to explore technologies such as generative AI, large language models, retrieval-augmented generation, AI agents, tool calling, multimodal AI, evaluation, observability and human oversight.
Importantly, these are project ideas for discussion, not finished project packages. Detailed implementation plans, datasets, technology choices and feasibility assessments will be developed for projects that attract sufficient interest.
What Makes These Projects AI-Native?
In a conventional AI application, a model may perform one specific task, such as prediction or classification.
In an AI-native application, intelligence becomes a central part of the system's workflow. The system may interpret a goal, gather relevant information, use software tools, generate possible actions, verify results and seek human approval.
The engineering challenge is therefore not just to obtain an AI-generated answer, but to build a system that produces useful, reliable and verifiable outcomes.
12 Proposed AI-Native Engineering Projects
AI Agent That Builds and Tests Software
Build an AI agent that converts software requirements into working code, executes tests, identifies errors and proposes fixes. The challenge is to make the agent reliable, secure and capable of verifying its own work.
AI Researcher That Verifies Its Own Claims
Develop an AI research assistant that searches documents, gathers evidence and produces referenced findings. The challenge is detecting unsupported claims, conflicting information and hallucinations.
AI Agent That Detects Prompt Injection
Create an AI security system that identifies malicious instructions hidden in documents, web pages or tool responses. The challenge is preventing attacks without blocking legitimate user requests.
AI Project Manager That Revises Its Own Plans
Develop an AI agent that converts project goals into tasks, schedules and milestones. The challenge is adapting plans when deadlines, resources or priorities change.
Multilingual Voice-to-Action AI Agent
Build a voice-enabled AI agent that understands requests in Indian languages and performs approved digital tasks. The challenge is handling accents, mixed languages, ambiguity and safe action execution.
AI Document-to-Workflow Engineer
Create an AI system that reads invoices, forms or technical documents and converts their contents into structured workflows. The challenge is extracting accurate information, detecting inconsistencies and preventing incorrect actions.
AI System That Evaluates Other AI Systems
Develop a platform that tests AI applications for accuracy, hallucinations, response quality, latency and cost. The challenge is designing meaningful evaluation criteria and detecting failures consistently.
Self-Improving AI Knowledge Retrieval System
Build a retrieval-augmented AI system that learns from unsuccessful searches and improves how it finds information. The challenge is demonstrating measurable improvements without introducing incorrect or irrelevant knowledge.
Multi-Agent Engineering Design Review
Create a team of specialized AI agents that examine an engineering design from technical, financial, safety and operational perspectives. The challenge is resolving conflicting recommendations and producing an evidence-based final assessment.
Natural Language to Verified SQL Agent
Develop an AI agent that translates everyday questions into database queries and explains the results. The challenge is generating correct queries while protecting sensitive data and preventing unsafe database operations.
Human-in-the-Loop AI Decision Agent
Build an AI system that recommends decisions, explains its supporting evidence and requests human approval before consequential actions. The challenge is balancing automation, accountability, uncertainty and human control.
Lateral AI Innovation Engine
Develop an AI system that challenges assumptions, introduces deliberate provocations and generates unconventional solutions to engineering problems. The challenge is moving beyond idea generation to identifying feasible concepts and designing practical experiments.
Designed for Two-Semester Engineering Projects
These ideas are proposed with the expectations of Indian engineering institutions in mind.
Each selected project would be developed into a structured learning and implementation programme that supports:
- Teams of three or four engineering students.
- Two semesters of development and evaluation.
- Meaningful individual engineering contributions.
- Accessible development tools and realistic computing costs.
- Identifiable datasets, knowledge sources or evaluation cases.
- A working software prototype and measurable results.
- Project documentation, demonstration and viva preparation.
A project would move into detailed development only after confirming that the necessary resources, data and evaluation methods are realistically available to students.
What Happens Next?
This initial list is being shared to invite reactions from engineering students, faculty members, project guides, academic institutions and industry professionals.
Based on the response, selected projects can be developed into complete engineering project packages covering:
- Industry problem and project objectives.
- Functional and technical requirements.
- System architecture and technology stack.
- Data, knowledge and evaluation resources.
- Semester-wise implementation milestones.
- Team roles and responsibilities.
- Testing, evaluation and success criteria.
- Final demonstration and project documentation.
An Invitation to Students and Faculty
Which of these 12 projects would you like to see developed further?
If you are an engineering student, faculty member or industry professional, share your preferred project number in the comments.
You are also welcome to suggest another AI-native engineering challenge that deserves to be included.
Your response will help prioritize the projects for detailed development.
From Learning AI to Engineering AI
The real opportunity for engineering students is to move beyond using AI tools and begin understanding how intelligent systems are actually engineered.
AIYING's proposed project lab aims to support that transition: from exploring possibilities to building, testing and demonstrating working AI systems.
Two semesters. One meaningful engineering challenge. A team of emerging AI engineers.
