Baglan Sharipov

AI Engineer

BCC

Builds and deploys LLM systems, including multi-agent and RAG applications, fine-tuning, and model serving infrastructure.

• Architecture and deployment of production-grade LLM solutions: open-source models (LLaMA 3, Qwen 2.5, Gemma 2, Mistral, Phi-3) and proprietary APIs (GPT-4o, Claude, Gemini); model selection based on task requirements and infrastructure constraints. • Built multi-agent systems with LangGraph and LangChain: specialized agents for planning, tool use, and memory management; RAG pipelines with hybrid search, re-ranking, and vector databases (Chroma, FAISS, Qdrant).

Data Scientist

Kcell

Developed predictive models, recommender systems, forecasting, and NLP tools, with model pipelines managed in Airflow.

• Developed and deployed predictive models for customer churn, fraud detection, lead generation, uplift modeling, and propensity-to-buy; recommender systems using collaborative, content-based, and hybrid filtering approaches. • Time series forecasting with (S)ARIMA, Prophet, LSTM, and Transformer-based models; built LLM-powered chatbots and NLP pipelines (classification, NER, sequence labeling) for customer-facing automation.

Data Scientist

KontOne

Applied predictive modeling and LLM agents to trading, and developed blockchain smart contracts for finance use cases.

• Portfolio optimization and predictive modeling in Python: multiclass classification, gradient boosting (XGBoost, LightGBM), decision trees, and deep learning for rebalancing and trading signal generation. • Integrated LLMs (GPT, LLaMA) with trading APIs using LangChain, semantic search, and vector databases to build autonomous agents capable of real-time market reasoning and strategy execution. • Developed smart contracts in Solidity and Go on Ethereum and IBM Hyperledger for ICOs, DeFi protocols, crypto hedge fund automation, and on-chain trade settlement.

Quantitative Trader / Data Analyst

Proprietary Trading

Designed and backtested trading strategies across several asset classes using order book, price, and volume data.

• Designed and backtested algorithmic strategies for stocks, futures, FX, and cryptocurrencies; modeled market microstructure including liquidity, slippage, and transaction costs. • Signal generation from order book depth, OHLC candlesticks, and volume data; strategy parameter optimization using GridSearch, Optuna, and Bayesian Optimization; real-time monitoring and performance logging.

MSc, Information Technologies

University College London