AI/ML Engineer

CallKaro AI | Onsite, Gurgaon

What Is an AI/ML Engineer?

An AI/ML Engineer is a hands-on technical role responsible for building, training, fine-tuning, evaluating, and deploying machine learning models that solve real business and product problems.

This role is not client-facing. The focus is on model development, experimentation, data pipelines, GPU-based training, fine-tuning workflows, and taking models from research or prototype stage into reliable production systems.

The ideal person should be comfortable working close to the model layer: understanding data, choosing the right architecture, training models efficiently, debugging performance issues, and improving model quality through experimentation.

What You Will Do

- Build and train machine learning, deep learning, and AI models for real-world product use cases.
- Work with GPUs for model training, fine-tuning, optimization, and experimentation.
- Fine-tune open-source or internal models using structured datasets, domain-specific data, and task-specific objectives.
- Design and improve ML pipelines for data preprocessing, feature engineering, training, evaluation, and deployment.
- Experiment with model architectures, hyperparameters, loss functions, embeddings, and evaluation methods.
- Work with large datasets, clean noisy data, and prepare high-quality training and validation sets.
- Evaluate model performance using the right metrics and identify where models are failing.
- Optimize models for accuracy, latency, cost, memory usage, and production reliability.




- Build inference workflows and integrate trained models into backend services or internal tools.
- Use frameworks such as PyTorch, TensorFlow, Hugging Face, Scikit-learn, or similar ML libraries.
- Document experiments, model decisions, training results, and deployment learnings clearly.
- Stay updated with practical developments in machine learning, generative AI, fine-tuning, and model optimization.

What We Look For

Must-Have
- Strong understanding of machine learning fundamentals, including supervised learning, unsupervised learning, model evaluation, overfitting, regularization, and optimization.
- Hands-on experience training ML or deep learning models, not just using APIs.
- Practical experience working with GPUs for training, fine-tuning, or model experimentation.
- Solid Python skills with libraries such as PyTorch, TensorFlow, Scikit-learn, NumPy, Pandas, or Hugging Face.
- Ability to prepare datasets, clean data, build training pipelines, and debug data quality issues.
- Understanding of neural networks, embeddings, transformers, fine-tuning, and modern AI model workflows.
- Ability to evaluate models properly using metrics, test sets, validation methods, and failure analysis.
- Comfort reading research papers, technical documentation,



and adapting ideas into working implementations.
- Strong problem-solving ability and patience for experimentation, iteration, and debugging.
- Ownership mindset: you care about whether the model actually works in production, not just whether it trains.

Good to Have

- Experience fine-tuning LLMs, vision models, speech models, recommendation models, or other deep learning systems.
- Experience with LoRA, QLoRA, PEFT, quantization, distillation, or other model optimization techniques.
- Exposure to MLOps tools, experiment tracking, model versioning, data versioning, or deployment pipelines.
- Experience with cloud GPU platforms, CUDA basics, Docker, Kubernetes, or distributed training.
- Familiarity with vector databases, RAG pipelines, semantic search, or embedding-based systems.
- Experience deploying ML models behind APIs or integrating them into production applications.
- Prior experience working in a startup, research lab, AI product team, or fast-moving engineering environment.

The Mindset We Hire For

We want someone who is deeply technical, practical, and experiment-driven. This role is for a person who enjoys making models work in the real world, not only discussing them theoretically.

You should be able to look at a problem, understand what data is available, choose a sensible modeling approach, train and evaluate it properly, and keep improving it until it becomes useful. We value people who are curious, rigorous, honest about model limitations, and willing to iterate until the system performs well in production.

📌 AI/ML Engineer (Gurugram)
🏢 CallKaro.AI
📍 Gurugram