LLM & Deep Learning Engineer — MedhasevCareers/AI Engineering# LLM & Deep Learning EngineerBuild and run production LLM systems: fine-tuning, evaluation, serving, and keeping them reliable.* Bengaluru (Remote)* Full-Time* 4+ years* Rs 18-24 LPAApply NowAll open roles## About the roleYou have shipped transformer-based systems before and want to own that work here. Day to day you will fine-tune models, improve retrieval quality, speed up inference, and put evaluation in place so regressions get caught early. Comfort reading papers and explaining trade-offs to the rest of the team matters.## Responsibilities* Fine-tune and adapt foundation models (SFT, LoRA/QLoRA/PEFT, preference methods when useful).* Design and improve RAG pipelines: chunking, embeddings, reranking, and grounded answers.* Improve inference cost and latency (quantization, batching, caching, GPU use).* Build evaluation sets and regression checks for quality and safety.* Handle model versioning, rollouts, monitoring, and basic incident response.* Work with product and software engineers to ship model-backed features.* Review designs, write runbooks, and help others on the team level up.## Required qualifications* Strong Python and solid software habits (tests, types, reviews)* PyTorch in real projects (training loops, debugging, multi-GPU familiarity)* Transformers: attention, decoding, and practical scaling trade-offs* Hugging Face (Transformers, Datasets, Accelerate / PEFT)* LLM fine-tuning in production: LoRA, QLoRA, PEFT, dataset curation* RAG with vector databases (FAISS, Pinecone, Weaviate, pgvector, or similar)* Model optimization and deployment (quantization, ONNX/TensorRT or similar)* Operating AI services in production (latency, cost, logs, failure modes)* Turning research papers into working code## Preferred skills* vLLM, TGI, TensorRT-LLM, Triton, or similar serving stacks* DPO, ORPO, RLHF, or related preference tuning* Multimodal, long-context, or tool-use pipelines* CUDA profiling and GPU memory tuning* Prior ownership of an ML or LLM platform area## CompensationRs 18-24 LPA₹18 LPA as specified by Medhasev. Final offer depends on interview performance.## Location & arrangementBengaluru, Karnataka, India (Remote)Remote-first with a Bengaluru base. Async-friendly collaboration across India time zones.## EducationBachelor's or Master's in Computer Science, AI, ML, or equivalent experience shipping deep learning systems## Benefits* Remote work from anywhere in India* Direct ownership of real product work* Close collaboration with founders and senior engineers* Learning budget for courses, papers, and tools* Flexible hours built around deep work## Application process* Submit your application with resume through this careers portal.* We review applications on a rolling basis and reply to shortlisted candidates.* Shortlisted candidates complete a practical technical screen.* Conversation with the team about craft, ownership, and fit.* Offer and onboarding for selected candidates.Apply for this roleOr email careers@medhasev.com