Data Scientist / Deep Learning Engineer (Neural Networks / Time Series / NLP)

Company: Spice IT ☑️Stack Neural networks on sequential data (Transformers, RNN, LSTM, GRU, etc.) + classic ML (scoring). Pytorch, Keras, Python, pandas, numpy, matplotlib, sklearn, lgbm, xgb, Hadoop, Greenplum.

☑️Responsibilities -Conducting R&D on training neural network models on near-raw sequential data. -Focus on building neural networks for assessing the creditworthiness of individuals at various stages of the client lifecycle (pre-approval, application review, recovery, bankruptcy). -Full cycle of work on the model: from ordering to production deployment. -Interaction with product verticals in formalizing business initiatives, defending results within the DS responsibility area.

☑️What we expect from you -At least 3-5 years of experience as a Data Scientist. -Experience in classic NLP (BERT) / transformers (focus only on LLM experience is not suitable). -Experience with sequences, time series. -Experience with recurrent neural networks (RNN, LSTM, GRU) - will be a plus. -Experience with large volumes of data (samples of 1+ million observations X 1000+ features). -Excellent knowledge of SQL and experience with relational DBMS and corporate data warehouses (Greenplum, Oracle, MSSQL, PostgreSQL). -Experience in fintech (credit bureaus, banks, MFIs, etc.) – will be a plus. ** ☑️What we are ready to offer** -Motivation system: fixed monthly salary + annual bonus (up to 10% of annual income). -Official employment according to the Labor Code of the Russian Federation. -The company is included in the register of accredited IT companies. -Remote work format (within the Russian Federation!). The office is located in Moscow. -Voluntary medical insurance with dental care (from the 1st working day + travel insurance). -Partial compensation for fitness / training / additional medical services (limit – 25 thousand rubles per year). -Additional days to annual leave (total 31 days per year). -Modern equipment (Windows, Lenovo ThinkPad). -Opportunities for growth within the company.

☑️How we conduct interviews

1. 25-30 minutes – conversation with a recruiter (video meeting).
2. 60 minutes – technical interview (video meeting) – discussion of experience; live coding: SQL tasks, Python task; block of questions on classic ML.
3. 60 minutes – second technical interview (video meeting) – theory and a couple of tasks on neural networks.
4. Document check 1-2 days (electronic questionnaire).
5. Offer, discussion of the start date. Contacts: recruiter contact