DHI Solutions

Job Title : Gen AI Expert

Location : Gurgaon (WFO)

Experience Level : 4+ years in AI/ML development (with hands-on GenAI exposure). Must have real time projects exposure.

About the Role :

We are seeking a highly skilled Generative AI Developer who thrives at the intersection of deep learning research and real-world application. This role requires hands-on experience in building and deploying Transformer-based architectures, working with contrastive learning techniques, and implementing robust training pipelines leveraging batch normalization and other optimization techniques. You will be contributing to the development of intelligent systems with advanced natural language and multi-modal understanding, generation, and retrieval capabilities.

Key Responsibilities :

- Design, develop, and optimize Transformer-based models (e.g., BERT, GPT, T5) for a range of generative tasks such as text summarization, Q&A, content creation, and code generation.

- Implement and experiment with deep learning techniques, especially focusing on batch normalization, dropout, residual connections, and attention mechanisms to improve model training and convergence.

- Build and train contrastive learning frameworks (e.g., SimCLR, CLIP, MoCo) for representation learning and fine-tuning pre-trained models on domain-specific data.

- Develop scalable information retrieval systems, integrating dense retrieval and vector[1]based semantic search using FAISS or similar technologies.

- Optimize model performance for inference using techniques such as quantization, distillation, and hardware-aware tuning (GPU/TPU-based training).

- Collaborate with data scientists and product teams to define and deliver user-focused GenAI solutions.

- Stay updated with state-of-the-art research in generative AI, deep learning, and retrieval-augmented generation (RAG), and incorporate best practices into development.

Required Skills and Qualifications :

- Strong hands-on experience with Transformer architectures and their practical implementations using frameworks like Hugging Face Transformers, PyTorch, or TensorFlow.

- Deep understanding of deep learning fundamentals, including backpropagation, batch/layer normalization, weight initialization, and regularization.

- Practical experience with contrastive learning paradigms and implementing self supervised learning strategies.

- Solid grasp of information retrieval systems, vector databases, embedding generation, and semantic search methodologies.

- Proficiency in Python and ML frameworks (e.g., PyTorch, TensorFlow, JAX).

- Experience with end-to-end model lifecycle : data preprocessing, model design, training, evaluation, and deployment.

- Ability to handle large datasets and distributed training.

- Knowledge of MLOps tools (e.g., MLflow, Weights & Biases) and cloud platforms (AWS/GCP/Azure).

Nice to Have :

- Experience in Retrieval-Augmented Generation (RAG) pipelines.

- Familiarity with LoRA, PEFT, or other fine-tuning methods for large language models.

- Prior contributions to open-source GenAI projects.

- Experience with multi-modal models (text + image/audio).