Job
Description
About MostEdge At MostEdge , our purpose is clear: Accelerate commerce and build sustainable, trusted experiences. With every byte of data, we strive to Protect Every Penny. Power Every Possibility. We empower retailers to make real-time, profitable decisions using cutting-edge AI , smart infrastructure, and operational excellence. Our platforms handle: hundreds of thousands of sales transactions/hour hundreds of vendor purchase invoices/hour few hundred product updates/day With systems built for 99.99999% uptime We are building an AI-native commerce engine , and language models are at the heart of this transformation. Role Overview We are looking for an AI/ML Expert with deep experience in training and deploying Large Language Models (LLMs) to power MostEdge's next-generation operations, cost intelligence, and customer analytics platform . You will be responsible for fine-tuning domain-specific models using internal structured and unstructured data (product catalogs, invoices, chats, documents), embedding real-time knowledge through RAG pipelines, and enabling AI-powered interfaces that drive search, reporting, insight generation, and operational recommendations. Scope & Accountability What You Will Own Fine-tune and deploy LLMs for product, vendor, and shopper-facing use cases. Design hybrid retrieval-augmented generation (RAG) pipelines with LangChain, FastAPI, and vector DBs (e.g., FAISS, Weaviate, Qdrant). Train models on internal datasets (sales, cost, product specs, invoices, support logs) using supervised fine-tuning and LoRA/QLoRA techniques. Orchestrate embedding pipelines, prompt tuning, and model evaluation across customer and field operations use cases. Deploy LLMs efficiently on RunPod, AWS, or GCP , optimizing for multi-GPU, low-latency inference . Collaborate with engineering and product teams to embed model outputs in dashboards, chat UIs, and retail systems. What Success Looks Like 90%+ accuracy on retrieval and reasoning tasks for product/vendor cost and invoice queries. <3s inference time across operational prompts, running on GPU-optimized containers. Full integration of LLMs with backend APIs, sales dashboards, and product portals. 75% reduction in manual effort across selected operational workflows. Skills & Experience Must-Have 5+ years in AI/ML , with 2+ years working on LLMs or transformer architectures . Proven experience training or fine-tuning Mistral, LLaMA, Falcon, or similar open-source LLMs . Strong command over LoRA, QLoRA, PEFT, RAG, embeddings, and quantized inference . Familiarity with LangChain, HuggingFace Transformers, FAISS/Qdrant , and FastAPI for LLM orchestration. Experience deploying models on RunPod, AWS, or GCP using Docker + Kubernetes. Proficient in Python , PyTorch , and data preprocessing (structured and unstructured). Experience with ETL pipelines , multi-modal data, and real-time data integration. Nice-to-Have Experience with retail, inventory, or customer analytics systems . Knowledge of semantic search, OCR post-processing, or auto-tagging pipelines . Exposure to multi-tenant environments and secure model isolation for enterprise use. How You Reflect Our Values Lead with Purpose : You empower smarter decisions with AI-first operations. Build Trust : You make model behavior explainable, dependable, and fair. Own the Outcome : You train and optimize end-to-end pipelines from data to insights. Win Together : You partner across engineering, ops, and customer success teams. Keep It Simple : You design intuitive models, prompts, and outputs that drive action—not confusion. Why Join MostEdge? Shape how AI transforms commerce and operations at scale . Be part of a mission-critical, high-velocity, AI-first company . Build LLMs with purpose—connecting frontline data to real-time results. Job Types: Full-time, Permanent Pay: ₹1,068,726.69 - ₹2,729,919.70 per year Benefits: Health insurance Life insurance Paid sick time Paid time off Provident Fund Schedule: Evening shift Morning shift US shift Supplemental Pay: Performance bonus Yearly bonus Work Location: In person Expected Start Date: 15/07/2025
