New opportunity

AI Engineer

CCTECH · Pune City, India

About this role

CCTECH is seeking an AI Engineer for its project delivery team in Pune City, India. This permanent role focuses on designing, developing and deploying enterprise-grade applications powered by generative AI and large language models (LLMs). The posting specifies 2–8 years of work experience and requires at least 2 years in AI/ML, with strong exposure to generative AI systems.

Responsibilities

Build LLM applications and develop prompt-engineering strategies to improve output accuracy, relevance and performance. Use cases include chatbots, document processing, summarization and knowledge assistants.
Implement preprocessing, chunking and embedding workflows for unstructured data; integrate generative AI models into APIs, microservices and enterprise applications.
Monitor and improve model responses, latency and cost efficiency. Define evaluation metrics for accuracy, hallucination and relevance; test, validate and iteratively improve generative AI pipelines.
Work with product managers, engineers, stakeholders, data engineering and DevOps teams to translate requirements into solutions and support scalable production deployment. Participate in agile planning, stand-ups and design and code reviews. Document prompts, workflows and design decisions while applying data-privacy and safe-AI practices.

Required qualifications and skills:

Proficiency in Python; experience with OpenAI, Azure OpenAI or similar APIs; and experience handling structured and unstructured data.
Experience with frameworks such as LangChain, LangGraph and LlamaIndex; modular AI solution design; RAG pipelines, vector databases and RAG evaluation; and prompt engineering, vector embeddings and context management.
Experience integrating AI through REST APIs, plus familiarity with Docker, CI/CD pipelines and Azure. Strong analytical, problem-solving, communication and collaboration skills; ownership and the ability to work in an agile environment.

Preferred qualifications include a bachelor's or master's degree in computer science, AI, data science or a related field; experience fine-tuning or adapting LLMs, building multi-agent or tool-augmented systems, and working with knowledge graphs or semantic search; and exposure to enterprise-scale AI applications or production deployments.

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

Skills for this role

PythonGenerative AILarge Language ModelsPrompt EngineeringOpenAI APIAzure OpenAILangChainLangGraphLlamaIndexRetrieval-Augmented GenerationVector DatabasesVector EmbeddingsData PreprocessingREST APIsDockerCI/CDAzureLLM EvaluationFine-TuningMulti-Agent SystemsKnowledge GraphsSemantic SearchModular CodingData Privacy

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

Build and integrate LLM applications
The role designs and develops GenAI applications for use cases such as chatbots, document processing, summarization, and knowledge assistants, then integrates models into APIs, microservices, or enterprise applications.
Develop reliable retrieval workflows
The posting calls for preprocessing, chunking, and embedding unstructured data, plus building RAG pipelines with vector databases. Prompt engineering, context management, and RAG evaluation are also required.
Evaluate and improve production behavior
The engineer is expected to define metrics for accuracy, hallucination, and relevance; test and refine pipelines; and monitor and debug responses, latency, and cost efficiency.
Deliver collaboratively and responsibly
The role translates stakeholder requirements into solutions, works with data engineering and DevOps on scalable production deployment, participates in agile reviews, and documents designs while applying data-privacy and safe-AI practices.

Likely challenges to think through

Balancing answer quality against latency and cost
Preparation hypothesis: improving relevance or reducing hallucinations may require more retrieval, context, or model calls, which can increase latency and cost. Prepare to explain how you would set evaluation metrics, compare alternatives, and monitor all three dimensions after deployment.
Making unstructured data useful for retrieval
Preparation hypothesis: chunking and embeddings can lose important context or retrieve irrelevant material, even when the LLM is capable. Prepare to discuss how you would test preprocessing and retrieval choices against representative questions before refining prompts.
Moving from a working prototype to a governed application
Preparation hypothesis: integrating an LLM into production requires balancing rapid iteration with API reliability, privacy, safe usage, and coordination across teams. Prepare to describe a modular integration approach, validation before release, and documentation of prompts and design decisions.

Practice your answers

Technical practice

How would you improve an LLM prompt for a knowledge assistant?

What to coverThe role calls for prompt strategies that improve accuracy, relevance and performance. Suggested preparation: explain how you would compare prompt versions against representative questions and examine failures before changing instructions or context.

How would you prepare unstructured documents for an LLM application?

What to coverThe posting names preprocessing, chunking and embedding workflows for unstructured data. Suggested preparation: discuss how document structure and chunk boundaries affect retrieval, and how you would check whether useful information survives preprocessing.

How would you design and evaluate a RAG pipeline?

What to coverThe role requires RAG pipelines, vector databases, embeddings and RAG evaluation. Suggested preparation: trace a query from retrieval through answer generation, then distinguish retrieval failures from answers that misuse retrieved context.

How would you evaluate a document summarization or chatbot feature?

What to coverThe posting calls for testing LLM outputs for accuracy, hallucination and relevance across use cases including summarization and chatbots. Suggested preparation: describe representative test cases, how you would assess each metric, and how findings would guide iteration.

How would you investigate poor responses, high latency or rising costs?

What to coverMonitoring and improving responses, latency and cost efficiency are stated responsibilities. Suggested preparation: explain what you would measure across retrieval, context and model calls, and how you would weigh quality against speed and cost.

How would you integrate an LLM workflow into an enterprise application?

What to coverThe role includes Python, LLM APIs, REST integration, microservices and modular design. Suggested preparation: outline component boundaries, request and response handling, and how you would test the integration.

How would you support a scalable production deployment?

What to coverThe posting mentions Docker, CI/CD and Azure familiarity, as well as collaboration with data engineering and DevOps. Suggested preparation: describe deployment checks and how you would detect and respond to pipeline or application failures without assuming a specific company setup.

How would you address privacy and safe AI usage in an LLM application?

What to coverThe role explicitly requires data privacy and responsible, safe AI practices. Suggested preparation: explain how you would review data sent to a model, handle sensitive content and validate outputs for the intended use case.

When might you adapt or fine-tune an LLM rather than revise the prompt or retrieval?

What to coverFine-tuning or adapting LLMs is listed as preferred experience, not a required duty. Suggested preparation: compare the approaches in terms of the failure they address, evaluation effort and maintenance cost.

When would a tool-augmented or multi-agent design be justified?

What to coverTool-augmented LLMs and multi-agent systems appear in the preferred qualifications. Suggested preparation: describe a task that might benefit from tools, and weigh that benefit against added complexity, failure modes and evaluation needs.

Behavioural practice

Tell me about a time you turned an unclear business need into a technical solution.

What to coverThe role involves translating requirements with product managers, engineers and stakeholders. Suggested preparation: use a real situation, your actions to clarify the need and the resulting decision or outcome.

Describe a time you coordinated a deployment across teams.

What to coverThe posting calls for working with data engineering and DevOps on scalable deployments and production integration. Suggested preparation: give a real example covering the handoffs, your actions and the result.

Tell me about a design or code review that changed your approach.

What to coverParticipation in design and code reviews is part of the role. Suggested preparation: describe the situation, how you handled the feedback and what changed as a result.

Describe a time you took ownership of a difficult system issue.

What to coverThe posting emphasizes problem-solving, ownership and delivery, alongside debugging AI responses. Suggested preparation: choose a real issue, explain your investigation and actions, and state the outcome without overstating your contribution.

Tell me about a time you documented a technical decision for others.

What to coverThe role calls for documenting prompts, workflows and system design decisions. Suggested preparation: explain the situation, what you documented and why, and how it helped collaborators or later work.

Sharpen your knowledge

Topics to prepare

LLM prompting and context management
Posting: Prompt strategies should improve output accuracy, relevance and performance; context management is required. Suggested revision exercise: explain how you would structure and revise a prompt for a document summarization task when the output omits important facts.
RAG and vector retrieval
Posting: RAG pipelines, vector databases, embeddings and RAG evaluation are required. Suggested revision exercise: diagram how an unstructured document moves through preprocessing, chunking, embedding, retrieval and answer generation; identify where irrelevant context could enter.
LLM evaluation and iteration
Posting: Define metrics for accuracy, hallucination and relevance, then test and improve pipelines. Suggested revision exercise: outline a small question-and-answer test set and explain how you would distinguish an unsupported answer from an irrelevant one.
Production integration and trade-offs
Posting: GenAI components are integrated into APIs, microservices and enterprise applications, with attention to response quality, latency, cost and scalable deployment. Suggested revision exercise: sketch an application boundary and explain what you would monitor and adjust if responses became slow or expensive.
Data privacy and safe AI
Posting: The role calls for data-privacy and safe-AI practices. Suggested revision exercise: walk through a document-processing use case and identify where sensitive input or generated output needs scrutiny; this is preparation, not a claim about CCTECH's specific controls.

Skills to revise

Python LLM API integration
Posting: Python proficiency and experience with OpenAI, Azure OpenAI or similar APIs are required. Suggested revision exercise: write a small Python function that sends a prompt, handles an API failure and returns a usable response.
Unstructured-data preparation
Posting: Implement preprocessing, chunking and embedding workflows. Suggested revision exercise: take a sample document, clean and split it into chunks, then explain how you would create embeddings and preserve enough context for retrieval.
Modular RAG implementation
Posting: Build RAG pipelines with vector databases using modular code; LangChain, LangGraph and LlamaIndex are named frameworks. Suggested revision exercise: implement or outline separate ingestion, retrieval and answer-generation components using a named framework you know.
Prompt and pipeline testing
Posting: Improve model responses through testing, validation and RAG evaluation. Suggested revision exercise: compare two prompts on the same document questions, record accuracy, hallucination and relevance observations, and document one justified change.
REST and deployment workflow
Posting: REST API integration is required, with familiarity in Docker, CI/CD and Azure. Suggested revision exercise: outline a REST endpoint for a knowledge assistant, then describe how you would package it with Docker and validate a change through a CI/CD pipeline.
Design communication and documentation
Posting: Collaborate across product, engineering, data engineering and DevOps; participate in agile planning and reviews; document prompts, workflows and design decisions. Suggested revision exercise: prepare a brief design note for a chatbot, including its requirements, component boundaries, evaluation plan and open questions for stakeholders.

Your preparation checklist

Map evidence to the required skills
Prepare a one-page list of examples you can discuss for Python, LLM APIs, structured and unstructured data, RAG, REST integration, and Docker, CI/CD and Azure familiarity. Mark gaps rather than inventing experience.
Prepare a GenAI project walkthrough
If you have a relevant project, outline the problem, your contribution, architecture, design decisions and measured results. Choose an example involving an LLM application or explain a comparable design exercise if you do not have one.
Sketch a document-based RAG design
As interview practice, draw a knowledge-assistant flow covering preprocessing, chunking, embeddings, vector retrieval, context management and response generation. Explain where modular Python code and a framework such as LangChain, LangGraph or LlamaIndex would fit.
Build an evaluation and improvement plan
Prepare a small example test set and describe how you would assess accuracy, hallucination and relevance. Include how you would iterate on prompts or retrieval and monitor response quality, latency and cost.
Explain a production integration approach
Outline how an LLM component could be exposed through a REST API and deployed with Docker, CI/CD and Azure. Identify integration, debugging and scaling decisions you would discuss with engineering, data engineering and DevOps teams.
Prepare governance and delivery examples
Write down how you would handle sensitive data, safe AI usage and documentation of prompts and design decisions. Have a concise example ready of translating stakeholder requirements into work for an agile team.
Draft questions for CCTECH
Ask which chatbot, document-processing, summarization or knowledge-assistant use cases the project delivery team is prioritizing; how it evaluates RAG quality; and what production integration and privacy constraints shape the work.

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…