Description
Our team focuses on tasks related to information extraction from unstructured content: documents, dialogues, texts of various kinds.
Our main objective is creating ready-to-use products powered by LLMs, as well as building self-service platforms for business users, where anyone can create their own data processing scenario (skill) in a no-code mode. We have recently launched a tool where users can chat with a document in copilot mode, enhancing their work efficiency.
We are looking for colleagues interested in NLP, who are excited about both RnD and building E2E systems for business process automation.
Additionally, we work on AI agents for solving complex multi-stage tasks involving information analysis from various sources. For this purpose, we are developing our own tools (SDK, a library of functions, benchmarks for agent evaluation).
Responsibilities
- Development of applied LLM technologies for information extraction and generative search (RAG) tasks
- Fine-tuning (LoRA) of multimodal large language models with a focus on the document domain
- Development of AI agents and multi-agent systems
- Organization and automation of the data annotation process (from data search and preparation to error analysis)
- Releasing new models into execution environments for our users.
Requirements
- Experience with LLMs, prompt engineering, fine-tuning transformer models
- Experience in ML development of one or more model types: Text classification, NER, QA
- Excellent knowledge of PyTorch, Numpy, Sklearn, Pandas, Python3, OOP, SOLID
- LLMOps: LangChain, LlamaIndex, experience with LLM tools.
WILL BE A PLUS:
- Strong GitHub profile
- Kaggle medals
- NLP/LLM publications at international conferences
- Participation in open-source LLM projects
- Experience in optimizing and accelerating models for production (pruning, quantization, ONNX/TensorRT)
- MLOps: Git, Docker, MLFlow/DVC/ClearML, Airflow
- Good knowledge of algorithms and data structures
Conditions
- Comfortable modern office - near Kutuzovskaya metro station
- Annual salary review, yearly bonus
- Corporate gym and relaxation areas
- Access to over 400 educational programs from SberUniversity for professional and career development
- Extended voluntary health insurance, preferential insurance for family, and corporate pension program
- Flexible mortgage discount equal to 1/3 of the Central Bank's key rate
- Free SberPrime+ subscription, discounts on partner company products
- Referral bonus for recommending friends to join the Sber team
- Corporate pension program.
