AI Engineer

Job Description: AI Engineer – Large Language Models (LLMs)

Location, Hadapsar, Pune

Overview

We are seeking a highly skilled AI Engineer with deep expertise in Large Language Models (LLMs) such as OpenAI GPT, Google Gemini, Anthropic Claude, Mistral, LLaMA, etc. The ideal candidate should have both theoretical knowledge of deep learning architectures and hands-on experience in training, fine-tuning, optimizing, and deploying custom LLMs at scale. This role requires a strong foundation in machine learning, distributed systems, and applied NLP, as well as experience with the end-to-end lifecycle of building proprietary language models.

⸻

Responsibilities

• LLM Research & Development

• Design, train, and fine-tune large-scale transformer-based models (GPT-like architectures, Gemini, etc.).

• Implement advanced techniques such as LoRA, PEFT, RLHF (Reinforcement Learning from Human Feedback), Constitutional AI, Mixture of Experts (MoE).

• Develop strategies for pre-training (on large datasets) and domain-specific fine-tuning.

• Infrastructure & Scaling

• Build and manage distributed training pipelines using frameworks such as PyTorch, TensorFlow, JAX, DeepSpeed, Megatron-LM.

• Optimize GPU/TPU utilization for multi-node, multi-GPU training.

• Deploy LLMs in cloud environments (AWS, GCP, Azure) and on Kubernetes/Dockerized microservices.

• Applied NLP & Product Integration

• Develop prompt engineering frameworks, RAG (Retrieval-Augmented Generation), vector database integrations (Pinecone, Weaviate, FAISS, Milvus).

• Implement LLMOps pipelines for monitoring, logging, and updating deployed models.

• Build evaluation frameworks (BLEU, ROUGE, perplexity, human eval, etc.) to measure model quality.

• Security, Ethics & Privacy

• Ensure LLM alignment with safety standards, bias reduction, hallucination minimization, and data privacy compliance (GDPR, SOC2).

• Research guardrails, toxicity filters, red-teaming strategies.

⸻

Required Technical Skills

• Deep Learning & NLP

• Strong knowledge of Transformers, Attention Mechanisms, LLM architectures.

• Experience with Hugging Face Transformers, LangChain, LlamaIndex.

• Understanding of tokenization strategies (BPE, SentencePiece, Unigram).

• Programming & Tools

• Proficiency in Python, PyTorch, TensorFlow, JAX.

• Experience with CUDA, NCCL, Triton, ONNX, TensorRT for performance optimization.

• Familiarity with C++ or Rust for low-level performance-critical modules.

• Data Engineering

• Expertise in large-scale data preprocessing, text cleaning, dataset curation.

• Knowledge of web scraping, document parsing, knowledge graphs, embeddings.

• Systems & Infrastructure

• Experience with distributed training frameworks (DeepSpeed, Horovod, Ray, Megatron-LM).

• Hands-on with MLOps/LLMOps tools (Kubeflow, MLflow, Weights & Biases, BentoML).

• Proficiency in cloud platforms (AWS Sagemaker, GCP Vertex AI, Azure AI).

⸻

Qualifications

• Bachelor's or Master's (PhD preferred) in Computer Science, AI, Machine Learning, or related field.

• 4+ years of hands-on experience in training and deploying large NLP models.

• Proven track record of building or fine-tuning LLMs (published papers, GitHub repos, or deployed systems).

• Strong understanding of probability, statistics, optimization, and reinforcement learning.

• Experience with real-world applications of LLMs (chatbots, copilots, knowledge assistants, search, summarization, code generation, etc.).

⸻

Preferred (Nice-to-Have) Skills

• Experience with multimodal models (text + image, text + speech).

• Familiarity with agentic AI frameworks (AutoGPT, CrewAI, BabyAGI).

• Contributions to open-source AI projects.

• Knowledge of federated learning & edge deployment of LLMs.

⸻

What We Offer

• Opportunity to build proprietary LLMs from scratch.

• Work alongside leading researchers and engineers in AI.

• Cutting-edge GPU/TPU infrastructure for experimentation.

• Competitive salary