Focus area: LLM Training & Evaluation

Location:
India, Chennai or Bengaluru preferred
· Hybrid

Experience:
6–10 years overall, including hands-on LLM fine-tuning Role Summary Own the training, fine-tuning and deployment of specialised language models for our agentic AI platform for airline operations. This is an applied, hands-on role: adapting and serving open-weight models in production.

Key Responsibilities
• Build and maintain training-data preparation and model-training pipelines.
• Fine-tune and adapt models using supervised fine-tuning, LoRA/QLoRA and other suitable techniques.
• Design model-level benchmarks and offline evaluations for the models you train; analyse failures and improve performance.
• Manage experiments, dataset and model versions, reproducibility and model documentation.
• Deploy and optimise models for quality, latency, throughput and cost.




• Work with Agent Engineering on model integration and with the Test & AI Evals Lead on production readiness. Required Skills & Experience
• Strong Python and PyTorch; practical experience with Hugging Face Transformers, Datasets and PEFT (or equivalent).
• Proven hands-on model training or fine-tuning experience, beyond prompting or consuming LLM APIs.
• Experience with dataset preparation, evaluation design, leakage prevention and experiment tracking.
• Sound understanding of LLM architectures, embeddings, RAG and model-adaptation trade-offs.
• Experience with GPU workloads, model serving, APIs, Docker and cloud infrastructure.
• Strong software engineering, troubleshooting and technical communication skills. Positive to Have
• Distributed training, preference optimisation, vLLM/Triton, MLOps, enterprise data privacy.

📌 Senior AI (Chennai)
🏢 dCortex
📍 Chennai