Padmi

Senior Data Scientist - Generative AI

IndiaPosted 1 month ago
Data Science And StatisticsSeniorFull Time; Regular
Apply at Tecnoprism

Opens the source posting on shine.com

Source description

About the role

View original

Job Description : Required Qualifications & Skills : Technical Skills : - Programming : Expert proficiency in Python and deep familiarity with asynchronous programming. - Deep Learning Frameworks : Extensive experience with PyTorch (preferred) or TensorFlow. - GenAI Stack : Hands-on experience with Hugging Face ecosystem (Transformers, Accelerate, PEFT, Diffusers) and LangChain or LlamaIndex. - Inference Optimization : Practical knowledge of quantization methods and high-throughput serving frameworks (e.g., vLLM). - Infrastructure : Proficiency with Docker, Kubernetes, and cloud platforms (AWS, GCP, or Azure) specifically for GPU-accelerated workloads. - Math & Theory : Strong foundation in linear algebra, probability, and the mathematical mechanics of Transformer architectures. Roles & Responsibilities : 1. Generative AI Development : - Work with various model architectures (Transformer-based, Diffusion models, etc.) to solve complex business problems. - Evaluate model performance using specialized metrics (Perplexity, BLEU, ROUGE, or custom LLM-as-a-judge frameworks). 2. AI Inference Engineering : - Optimize model architectures for efficient deployment, focusing on reducing latency and memory footprint. - Implement inference optimization techniques including Quantization (GGUF, AWQ, GPTQ), Pruning, and Knowledge Distillation. - Utilize high-performance inference engines and frameworks such as TensorRT, vLLM, ONNX Runtime, or DeepSpeed-Inference. - Design scalable serving infrastructures using tools like NVIDIA Triton Inference Server, BentoML, or Ray Serve. - Optimize GPU utilization and memory management for large-scale model serving. 3. Data & Model Lifecycle : - Curate and preprocess large-scale datasets for pre-training, fine-tuning, and RLHF (Reinforcement Learning from Human Feedback). - Build automated pipelines for model evaluation and continuous monitoring of production models (detecting drift or hallucinations). - Collaborate with ML Engineers to integrate models into scalable microservices. Education : - UG : Any Graduate .

One address, no account. We’ll tell you when matching roles go live.

More at Tecnoprism

Related open roles

View all roles