New opportunity

Lead - AI

SoftTech Engineers Limited · Pune, India

About this role

SoftTech Engineers Limited seeks an AI engineering leader in Pune to define the technology vision and roadmap for AI-powered Metaverse products and lead their development from concept through deployment. This is a permanent role in the architecture and construction industry; the stated salary is “As per Market Standard.”

Responsibilities

Architect and deliver LLM-powered applications, AI agents, copilots and intelligent assistants; build scalable RAG systems connected to vector databases and enterprise knowledge repositories.
Develop and optimize machine learning and deep learning models, and lead research and implementation of Transformer architectures.
Work with Product, Engineering, UX, BIM, CAD/CAE, Digital Twin and Metaverse teams. Establish AI governance, evaluation, security and responsible AI practices; drive cloud-native deployment and MLOps.
Lead and mentor AI engineering and data science teams, manage stakeholders, and communicate with executives, clients and technical teams. Success measures include product delivery, scalable LLM and RAG deployments, innovation, team growth and customer satisfaction.

Required qualifications

At least 6 years of software engineering and product development experience, including at least 3 years of hands-on machine learning and AI product development.
Expertise in generative AI and LLMs, RAG, Transformers and attention mechanisms; experience with PyTorch or TensorFlow, NLP, semantic search, embeddings, model training, fine-tuning, evaluation and optimization.
Python proficiency; experience with AWS, Azure or GCP, CI/CD, model monitoring and AI deployment pipelines. Ability to use prompt engineering for code writing and data extraction, and apply privacy, security and ethical data guidelines.
Product ownership, stakeholder management and experience leading large engineering teams delivering enterprise-scale products. Bachelor’s or Master’s degree in Computer Science, Artificial Intelligence, Machine Learning, Data Science or a related field.

Preferred qualifications

Experience with Metaverse, Digital Twin, XR/VR/AR, CAD, CAE, BIM or construction technology; AI agents and autonomous workflows; Graph RAG, knowledge graphs, agentic or multimodal AI; computer vision, spatial computing, Unity, Unreal Engine, distributed systems or high-performance computing. AI research publications, patents or contributions are also valued.

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

Skills for this role

AI product developmentAI strategyTechnical leadershipGenerative AILarge Language ModelsRetrieval-Augmented GenerationMachine LearningDeep LearningTransformersAI agentsPrompt engineeringPythonPyTorchTensorFlowNatural Language ProcessingSemantic searchEmbeddingsVector databasesKnowledge retrievalModel trainingModel fine-tuningModel evaluationModel optimizationMLOpsCI/CDModel monitoringAWSAzureGoogle Cloud PlatformAI governance intrinsics?

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Set the AI product direction
The role owns the AI technology vision and roadmap for Metaverse products and leads development from concept to deployment. Prepare to explain how you would connect technical choices to business objectives and product impact.
Deliver LLM and RAG products
The job calls for LLM applications, agents, copilots and assistants, plus scalable RAG connected to vector databases and enterprise knowledge repositories. Prepare to discuss architecture, retrieval quality and evaluation decisions.
Turn models into reliable deployments
Responsibilities include developing and optimizing ML models, leading Transformer implementation, and driving cloud-native deployment and MLOps. Prepare an end-to-end example that distinguishes model work from deployment, monitoring and iteration.
Build governance into delivery
The role must establish model evaluation, security and responsible AI practices; the requirements also cite privacy and ethical data guidelines. Prepare to describe how those controls would be applied throughout development rather than added at release.
Lead across disciplines
The lead mentors AI and data science teams and works with Product, Engineering, UX, BIM, CAD/CAE, Digital Twin and Metaverse teams. Prepare to show how you would make decisions and communicate trade-offs to technical teams, executives and clients.

Likely challenges to think through

Connecting AI to immersive products
Preparation hypothesis: integrating AI with Metaverse and construction-domain workflows may require balancing technical ambition against a usable product. Prepare to explain how you would validate a use case with UX and domain teams before committing to an architecture.
Balancing retrieval quality and scale
Preparation hypothesis: enterprise RAG may involve trade-offs among answer quality, latency, operating cost and access to sensitive knowledge. Prepare a concrete approach to retrieval evaluation, permission-aware design and monitoring.
Balancing innovation with delivery controls
Preparation hypothesis: research into Transformers and emerging AI approaches may compete with the need to ship and govern products. Prepare to describe criteria for choosing experiments, measuring value and deciding when a model is ready for deployment.

Practice your answers

Technical practice

How would you set the architecture and roadmap for an AI-powered Metaverse product?

What to coverThe role owns AI technology direction and delivery from concept to deployment. Suggested preparation: explain how you would sequence use cases, validate product value, and weigh integration complexity against scalability.

How would you design an LLM-powered copilot or AI agent for an enterprise product?

What to coverThe posting calls for LLM applications, agents, copilots, and intelligent assistants. Suggested preparation: discuss task boundaries, tool access, human oversight, and how you would test whether an agent adds value over a simpler assistant.

How would you build RAG over enterprise knowledge repositories?

What to coverThe role calls for scalable RAG connected to vector databases and enterprise knowledge. Suggested preparation: cover ingestion, access-aware retrieval, embeddings, grounding, and trade-offs between retrieval quality, latency, and cost.

How would you diagnose poor answers from a semantic-search or RAG system?

What to coverSemantic search, embeddings, knowledge retrieval, and model evaluation are stated skills. Suggested preparation: explain how you would distinguish missing or stale content, retrieval errors, and generation errors, then test changes against representative queries.

When would you fine-tune a Transformer model rather than use prompting or retrieval?

What to coverThe posting calls for Transformer and attention expertise, model training, fine-tuning, and experience with PyTorch or TensorFlow. Suggested preparation: explain the capabilities and limits of each approach, along with data, compute, maintenance, and evaluation trade-offs.

How would you evaluate and optimize an AI model before product release?

What to coverModel evaluation, optimization, and real-world business applications are explicit responsibilities. Suggested preparation: describe task-relevant test sets, quality and failure measures, and how latency or resource constraints would affect release decisions.

How would you establish privacy, security, and responsible-AI controls?

What to coverThe role includes AI governance and requires attention to privacy laws, security protocols, and ethical data guidelines. Suggested preparation: discuss data access, sensitive-information handling, misuse testing, and who reviews or responds to identified risks.

How would you deploy and monitor an AI service at enterprise scale?

What to coverCloud-native deployment, CI/CD, MLOps, model monitoring, and experience with AWS, Azure, or GCP are stated. Suggested preparation: outline a deployment pipeline, rollback approach, operational monitoring, and how you would investigate quality or performance regressions.

How would you use prompt engineering for code writing and data extraction?

What to coverThe posting explicitly names these two uses of LLM tools and requires strong Python skills. Suggested preparation: explain how you would specify outputs, validate extracted data or generated code, and handle unreliable results without treating model output as verified.

How would you work with BIM, CAD/CAE, and Digital Twin teams on an AI feature?

What to coverCross-team collaboration is a responsibility; experience in these domains and multimodal AI is listed as good to have. Suggested preparation: describe how you would establish data meaning, integration boundaries, and success criteria with domain specialists rather than assume a particular Metaverse implementation.

Behavioural practice

Tell me about a time you shaped a technical roadmap around a product goal.

What to coverDefining AI strategy aligned with business objectives is a leadership expectation. Suggested preparation: use a real situation, your prioritization and trade-offs, and the resulting product or delivery outcome.

Describe a time you aligned teams with different technical priorities.

What to coverThe role works with Product, Engineering, UX, and specialist domain teams. Suggested preparation: give a real situation, how you resolved differences in requirements or ownership, and what changed as a result.

Tell me about a time you mentored engineers while delivering a complex product.

What to coverLeading and growing AI engineering and data science teams is central to the role. Suggested preparation: explain the situation, your specific coaching and delegation actions, and the effects on team capability and delivery.

Describe a time you communicated an AI trade-off to executives or clients.

What to coverThe posting expects communication with executive leadership, clients, and technical teams. Suggested preparation: describe the audience and decision, how you conveyed uncertainty or risk, and the decision or outcome.

Tell me about a time a product or model fell short of expectations.

What to coverProduct delivery, model evaluation, engineering excellence, and customer satisfaction are stated priorities. Suggested preparation: use a real example to show how you identified the issue, coordinated a response, and measured the result without overstating success.

Sharpen your knowledge

Topics to prepare

Transformer and LLM fundamentals
Review attention mechanisms and how training, fine-tuning, evaluation, and optimization differ. Exercise: explain when you would fine-tune an LLM rather than use retrieval or prompt changes. The posting explicitly requires Transformer expertise and model-development experience.
RAG architecture and retrieval quality
Diagram a system connecting enterprise knowledge repositories, embeddings, semantic search, a vector database, and an LLM. Exercise: identify where retrieval can fail and how you would evaluate answer quality. Scalable RAG is a stated responsibility.
AI product strategy and governance
Exercise: draft a one-page AI roadmap with delivery milestones, evaluation criteria, privacy and security controls, and responsible-AI checks. The role owns AI strategy and product delivery and is expected to establish governance practices.
Metaverse and construction-domain integration
Exercise: sketch how an AI assistant could interact with a Digital Twin or BIM workflow, noting assumptions and cross-team dependencies. Metaverse products and collaboration with BIM, CAD/CAE, and Digital Twin teams are stated; prior domain experience is preferred, not required.

Skills to revise

Build and evaluate an LLM application
In Python, implement a small assistant and compare prompts and outputs against a defined test set. Practice prompt engineering for code writing or data extraction, both explicitly mentioned in the posting.
Implement a RAG prototype
Use sample documents to build an embedding and vector-database retrieval flow, then test whether answers are supported by retrieved passages. This rehearses the required hands-on RAG, semantic-search, and knowledge-retrieval work.
Train or fine-tune a model
Using PyTorch or TensorFlow, run a small training or fine-tuning exercise, record evaluation results, and explain an optimization decision. The posting calls for framework experience and model training, evaluation, and optimization.
Plan a deployment pipeline
Sketch a cloud deployment using AWS, Azure, or GCP, including CI/CD, model monitoring, and a response to degraded results. These are explicitly listed deployment and MLOps requirements.
Communicate a delivery trade-off
Prepare a short briefing for product, engineering, and executive audiences comparing two approaches to an AI feature, including delivery risks and evaluation criteria. The lead role requires product ownership, stakeholder management, and communication across technical and executive teams.

Your preparation checklist

Map the stated requirements to evidence
The posting specifies 6+ years in software engineering and product development, 3+ years hands-on in ML and AI product development, and a relevant degree. Prepare a one-page evidence map with dates, your role, and specific examples; leave gaps explicit.
Prepare an end-to-end product walkthrough
The role leads AI products from concept through deployment. Choose an actual project, if available, and outline the problem, roadmap decisions, architecture, your ownership, stakeholder trade-offs, delivery outcome, and measurable impact.
Practice a RAG architecture design
The posting calls for scalable RAG connected to vector databases and enterprise knowledge repositories. Sketch an LLM assistant design covering ingestion, embeddings, retrieval, access controls, evaluation, and scaling; label it as a proposed design, not company architecture.
Build a model-development deep dive
The role requires Transformers, attention mechanisms, Python, PyTorch or TensorFlow, and model training, fine-tuning, evaluation, and optimization. Prepare one real example showing technical choices, experiments, results, and limitations.
Show how deployment and governance work together
The posting names cloud deployment, MLOps, CI/CD, monitoring, privacy, security, and responsible AI. Prepare a deployment-and-evaluation diagram or checklist, distinguishing controls you have implemented from those you would propose.
Draft a Metaverse AI roadmap discussion
The role works with Product, UX, BIM, CAD/CAE, Digital Twin, and Metaverse teams. As a preparation exercise, outline how you would prioritize an AI use case, validate it with those teams, and define delivery and customer-impact measures; treat domain technologies listed as preferred, not mandatory.
Bring focused questions for the employer
Ask which AI-powered Metaverse products are priorities, what enterprise knowledge sources and evaluation measures are in use, and how product ownership and cross-team decisions are divided. These are questions to clarify the role, not assumptions about current systems.

Availability last checked: 5 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 India. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…