Job Title

Dear All, We are seeking a highly capableMachine Learning Engineerwith deep expertise in fine-tuningLarge Language Models (LLMs)andVision-Language Models (VLMs)for intelligent document processing. This role requires strong knowledge ofPEFT techniques(LoRA, QLoRA),quantization,transformer architectures,prompt engineering, and orchestration frameworks likeLangChain. You’ll work on building and scaling end-to-end document processing workflows using both open-source and commercial models (OpenAI, Google, etc.), with an emphasis on performance, reliability, and observability.Key Responsibilities: Fine-tune and optimize open-source and commercialLLMs/VLMs(e.G., LLaMA,Cohere, Gemini, GPT-4) for structured and unstructureddocument processingtasks. Apply advancedPEFT techniques(LoRA, QLoRA) andmodel quantizationto enable efficient deployment and experimentation. Design LLM-baseddocument intelligence pipelinesfor tasks like OCR extraction, entity recognition, key-value pairing, summarization, and layout understanding. Develop and manageprompting techniques(zero-shot, few-shot, chain-of-thought, self-consistency) tailored to document use-cases. ImplementLangChain-based workflows integrating tools, agents, and vector stores for RAG-style processing. Monitor experiments and production models usingWeights & Biases (W&B)or similar ML observability tools. Work withOpenAI (GPT series),Google PaLM / Gemini, and other LLM/VLM APIs for hybrid system design. Collaborate with cross-functional teams to deliver scalable, production-ready ML systems and continuously improve model performance. Build reusable, well-documented code and maintain a high standard of reproducibility and traceability. Performance evaluation of LLM and vLLM models for optimizing accuracy, latency.Required Skills & Experience: Hands-on experience withtransformer architecturesand libraries like HuggingFace Transformers. Deep knowledge offine-tuningstrategies for large models, includingLoRA,QLoRA, and otherPEFTapproaches. Experience inprompt engineeringand developing advanced prompting strategies. Familiarity withLangChain, vector databases (e.G., FAISS, Pinecone), and tool/agent orchestration. Strong applied knowledge ofOpenAI,Google (Gemini/PaLM), and other foundational LLM/VLM APIs. Proficiency inmodel training, tracking, and monitoringusing tools likeWeights & Biases (W&B). Solid understanding ofdeep learning,machine learning,natural language processing, andcomputer visionconcepts. Experience working withdocument AImodels (e.G., LayoutLM, Donut, Pix2Struct) and OCR tools (Tesseract, EasyOCR, etc.). Proficient inPython,PyTorch, and related ML tooling.Nice-to-Have: Experience withmulti-modal architecturesfor document + image/text processing. Knowledge ofRAG systems,embedding models, and custom vector store integrations. Experience in deploying ML models viaFastAPI,Triton, or similar frameworks. Contributions to open-source AI tools or model repositories. Exposure toMLOps,CI/CD pipelines, and data versioning.Qualifications: Bachelor’s or Master’s degree in Computer Science, Artificial Intelligence, Machine Learning, or a related field.Why Join Us? Work on cutting-edge GenAI and Document AI use-cases. Collaborate in a fast-paced, research-driven environment. Flexible work arrangements and growth-focused culture. Opportunity to shape real-world applications of LLMs and VLMs.Industry Software DevelopmentEmployment Type Full-time