New opportunity

Senior Machine Learning Engineer - NLP (Bengaluru/Noida)

Level AI · Bengaluru, India

About this role

Level AI is a Series C startup revolutionizing customer engagement through an AI-native platform that leverages Large Language Models to extract insights from customer interactions. As a Senior Machine Learning Engineer, you will play a high-impact role in shaping the future of AI-driven enterprise applications. Responsibilities include innovating with cutting-edge machine learning techniques to build data-driven solutions, working on NLP problems such as voice agents, agentic applications, text classification, entity extraction, and summarization. You will collaborate with cross-functional teams to integrate AI solutions, optimize models for performance and scalability, and build, deploy, and own production NLP pipelines. Additionally, you will implement and evaluate reasoning, planning, and memory modules for agents, establish evaluation metrics, and maintain post-deployment monitoring. Required qualifications include a B.Tech/M.Tech/PhD in computer science or mathematics-related fields from tier-1 institutes and 3+ years of industry experience in Machine Learning and NLP. Strong coding skills in Python and PyTorch are essential, along with familiarity with Transformers, LangChain, and LangGraph. You must have practical experience in NLP areas like text classification, entity tagging, information retrieval, and question-answering, as well as hands-on experience with Transformer-based models (BERT, Llama, etc.) and LLM training/inference optimization. Experience with ML/Deep Learning deployments using Docker, Kubernetes, and REST APIs is required. Desirable skills include knowledge of cloud platforms (AWS/Azure/GCP), multimodal models, and real-time streaming tools like Kafka. Bonus points are awarded for experience with open-source LLMs, vector search, RAG, prompt engineering, and contributions to AI projects.

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

Skills for this role

PythonPyTorchTransformersLangChainLangGraphNLPMachine LearningDeep LearningBERTLlamaQwenGemmaDeepSeekLLMREST APIDockerKubernetesAWSAzureGCPMultimodal modelsKafkaPub/SubVector searchRAGPrompt engineeringFAISSPinecone

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

Production-Grade NLP Pipeline Ownership
The role requires end-to-end responsibility for building, deploying, and maintaining scalable NLP pipelines. You must demonstrate proficiency in moving models from research to production using Docker and Kubernetes, while ensuring long-term reliability through post-deployment monitoring and continual learning.
LLM Integration and Agentic Workflows
A core priority is implementing reasoning, planning, and memory modules for AI agents. You will need to apply frameworks like LangChain and LangGraph to build functional agentic applications that solve complex customer engagement problems.
Model Optimization and Efficiency
You are expected to optimize existing models for performance, scalability, and efficiency. This involves deep knowledge of LLM training concepts, inference optimization techniques, and managing resource usage on GPUs to ensure real-time responsiveness.

Likely challenges to think through

Balancing Innovation with Production Stability
As a startup, the need to integrate cutting-edge SOTA techniques often conflicts with the requirement for stable, scalable production systems. Prepare by discussing how you prioritize technical debt and choose between experimental model architectures versus proven, stable deployments.
Establishing Rigorous Evaluation Metrics
Defining success for generative and agentic AI is inherently difficult. You will face the challenge of proposing and establishing benchmarks for subjective tasks like summarization or reasoning. Prepare by articulating your methodology for creating robust evaluation frameworks that correlate with real-world customer engagement outcomes.

Practice your answers

Technical practice

LLM Inference Optimization Techniques

What to coverDiscuss strategies for optimizing LLM inference, such as quantization, pruning, or speculative decoding. Explain how you would balance latency and throughput requirements in a production environment.

Designing Agentic Workflows with LangGraph

What to coverDescribe how you would structure an agentic application using LangGraph to handle complex reasoning tasks. Focus on how you manage state, loops, and error handling within the graph architecture.

RAG Pipeline Evaluation and Benchmarking

What to coverExplain your approach to evaluating a Retrieval-Augmented Generation (RAG) system. Detail the metrics you would use to assess both the retrieval component and the final generation quality.

Productionizing NLP Pipelines with Kubernetes

What to coverDescribe the challenges of deploying Transformer-based models in a Kubernetes environment. Address how you handle resource allocation, scaling, and model versioning for high-availability services.

Implementing Memory Modules for AI Agents

What to coverHow would you design a memory module for a long-running customer service agent? Discuss the trade-offs between short-term context windows and long-term vector-based storage.

Fine-tuning vs. Prompt Engineering

What to coverWhen would you choose to fine-tune a model like Llama versus using advanced prompt engineering or RAG? Explain the decision criteria based on data availability, cost, and performance requirements.

Handling Real-time Data with Kafka

What to coverExplain how you would integrate Kafka into an NLP pipeline for processing streaming customer interaction data. Focus on ensuring data consistency and low-latency ingestion for downstream inference.

Post-deployment Monitoring for NLP Models

What to coverWhat specific metrics and drift detection methods would you implement to monitor a production NLP model? Explain how you would trigger retraining or fine-tuning based on these observations.

Behavioural practice

Collaborating with Cross-functional Teams

What to coverDescribe a situation where you had to explain a complex machine learning trade-off to non-technical stakeholders. How did you ensure alignment on the project goals and technical approach?

Navigating Technical Ambiguity

What to coverShare an example of a time you were tasked with solving an ambiguous NLP problem with no clear SOTA solution. How did you approach the research, experimentation, and final implementation?

Prioritizing Technical Debt vs. Innovation

What to coverTell me about a time you had to balance the need for rapid feature delivery with the necessity of maintaining scalable, clean code. How did you decide what to refactor and what to defer?

Learning and Applying New AI Architectures

What to coverThe AI field moves rapidly; describe a recent instance where you had to quickly learn and implement a new model architecture or library. What was your process for mastering the technology and integrating it into your work?

Sharpen your knowledge

Topics to prepare

Agentic Workflow Architecture
The role requires implementing reasoning, planning, and memory modules for agents. Review the conceptual frameworks for agentic loops, specifically how memory (short-term vs. long-term) and planning (chain-of-thought or tree-of-thought) are integrated into LLM applications.
LLM Inference Optimization
Focus on techniques to improve model performance and scalability, such as quantization, speculative decoding, and efficient KV caching. This is critical for the requirement to own production NLP pipelines and optimize models for efficiency.
Evaluation Metrics for Generative AI
Study methodologies for establishing benchmarks in NLP tasks like summarization and question-answering. Research both automated metrics (e.g., ROUGE, BERTScore) and model-based evaluation frameworks to ensure production-grade quality.

Skills to revise

LangChain and LangGraph Implementation
Practice building multi-step agent workflows using LangGraph. Focus on managing state transitions and cyclic graphs, which are essential for the agentic applications mentioned in the job description.
Production Deployment with Docker and Kubernetes
Create a containerized deployment pipeline for a PyTorch-based model. Ensure you can expose the model via a REST API and manage its lifecycle within a Kubernetes cluster, reflecting the requirement to own production pipelines.
RAG and Vector Search Integration
Develop a retrieval-augmented generation pipeline using a vector database like FAISS or Pinecone. This exercise should demonstrate your ability to connect external knowledge bases to an LLM for improved question-answering accuracy.

Your preparation checklist

Prepare a technical project walkthrough on LLM-based agents
Develop a concise presentation or document detailing a project where you implemented an agentic application using LangChain or LangGraph. Focus on how you handled reasoning, planning, or memory modules, and be prepared to discuss the specific challenges of deploying these in a production environment.
Review production-grade NLP deployment architectures
Prepare to discuss your experience building and owning end-to-end NLP pipelines. Focus on your proficiency with Docker and Kubernetes for model serving, and explain how you have optimized model inference for performance, scalability, and efficiency in previous roles.
Synthesize your experience with RAG and vector search
Create a summary of your practical experience with retrieval-augmented generation (RAG) and vector databases like FAISS or Pinecone. Be ready to explain your approach to information retrieval, entity tagging, and how you established evaluation metrics for these systems.
Formulate questions about Level AI's agentic infrastructure
Prepare thoughtful questions regarding their current stack for reasoning and memory modules, their approach to post-deployment monitoring for LLMs, and how they handle the challenges of real-time streaming data integration using tools like Kafka.

Availability last checked: 18 Oct 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…