Artificial Intelligence Researcher

Skills : Applied Research , Architect and Innovation

Experience : 5 - 12 yrs

Location : Bangalore and Hyderabad

Key Responsibilities:

• Research & Innovation:

Conduct applied research in generative AI, foundation models, NLP, computer vision, and multimodal AI. Stay abreast of the latest publications and open-source advancements.

• Model Development:

Fine-tune, evaluate, and optimize large language models (LLMs), transformers, and other generative models for specific business and product use cases.

• Prototyping & Experimentation:

Build proof-of-concepts and experimental systems that demonstrate the potential of GenAI across domains such as content generation, summarization, synthetic data, agent systems, etc.

• Data & Evaluation Pipelines:

Design robust data pipelines, evaluation metrics, and benchmarking systems to validate model performance, safety, and bias.

• Collaboration:

Work with cross-functional teams including product managers, ML engineers, and data scientists to translate research into production-grade systems.

• Open Source & IP Contribution:

Publish findings in peer-reviewed venues, contribute to open-source projects, or generate intellectual property relevant to the business.

Required Qualifications:

• 5–7 years of experience in machine learning or applied AI roles, with at least 2–3 years working on generative models or related research.
• Strong foundation in deep learning frameworks such as PyTorch, TensorFlow, or JAX.
• Experience with LLMs (e.g., GPT, LLaMA, Claude), diffusion models, or vision-language models.
• Proficient in Python and ML tools/libraries such as Hugging Face Transformers, LangChain, or similar.
• Understanding of responsible AI practices, bias mitigation, and model explainability.
• Master’s or PhD in Computer Science, Machine Learning, Mathematics, or related fields.

Preferred Qualifications:

• Experience with open-source LLMs or fine-tuning techniques like LoRA, PEFT, RLHF, etc.
• Knowledge of MLOps practices and deployment of models in production (e.g., via Kubernetes, Ray, Triton).