Ilya Kipnis

Ilya Kipnis

Mentor
Rising Codementor
US$15.00
For every 15 mins
ABOUT ME
Senior AI/ML Engineer & Data Scientist
Senior AI/ML Engineer & Data Scientist

I'm a Senior AI/ML Engineer and Data Scientist with 12+ years of experience designing, building, and deploying machine learning models, AI-powered applications, and data-driven systems in production. My expertise spans Python, R, SQL, predictive modeling, quantitative research, generative AI, LLMs, and production-grade AI/ML architectures.

I've worked across finance, energy, and private equity, including BP, Bank of America, Deutsche Bank, and The Carlyle Group. At BP, I developed pricing algorithms and Shiny dashboards processing tens of millions of records, contributing to approximately $30M in savings. At The Carlyle Group, I re-engineered a Snowflake analytics pipeline covering 50+ financial metrics, reducing processing time from days to hours. I've also developed forecasting and portfolio allocation models for live strategies managing over $200M, reducing turnover by 60%.

Beyond traditional machine learning and quantitative modeling, I build modern AI solutions involving LLMs, retrieval-augmented generation (RAG), AI agents, NLP, semantic search, and intelligent data applications. I work across the full development lifecycle, from data engineering and model development to API integration, cloud deployment, and production optimization.

I've published 30+ open-source models, authored an SSRN paper ranked in the top 10% by downloads, and speak annually at R/Finance. I enjoy solving complex technical problems and turning research, data, and emerging AI technologies into reliable systems that deliver measurable business value.

English
Central Time (US & Canada) (-06:00)
Joined October 2026
EXPERTISE
8 years experience
10 years experience
10 years experience
10 years experience
8 years experience
10 years experience
10 years experience

REVIEWS FROM CLIENTS

Ilya's profile has been carefully vetted and approved as a Codementor. Connect with Ilya now, and leave a review for them once you're done!
EMPLOYMENTS
Data Scientist & Strategy Developer
Independent Quantitative Research
2014-01-01-Present
  • Designed and developed a Python library for portfolio backtesting and performance analysis, porting R functions ...
  • Designed and developed a Python library for portfolio backtesting and performance analysis, porting R functions to calculate portfolio drift, turnover, return attribution, Value at Risk (VaR), Conditional Value at Risk (CVaR), and drawdowns.
  • Published 30+ open-source quantitative models in Python and R covering asset allocation, equity signals, volatility modeling, and systematic investment strategies, with an emphasis on reproducibility, extensibility, and analytical rigor.
  • Developed, evaluated, and live-traded systematic volatility models for exchange-traded notes (ETNs), achieving approximately 100% returns in March 2020.
  • Created the Kipnis Defensive Adaptive strategy, combining momentum signals, minimum-variance optimization, and defensive indicators. Conducted parameter sensitivity analysis, portfolio optimization, and rebalancing tests to evaluate strategy robustness. The strategy is implemented on AllocateSmartly.
  • Applied statistical modeling, time-series analysis, and quantitative forecasting techniques to evaluate market behavior, identify investment signals, and assess portfolio risk.
  • Co-authored Momentum and Markowitz: A Golden Combination, an SSRN research paper ranked in the top 10% by downloads. The research was adopted by The Milwaukee Company.
  • Presented quantitative research and investment modeling work annually at the R/Finance conference.
Python
R
Pandas
View more
Python
R
Pandas
Machine Learning
Statistics
Data Visualization
Time Series Forecasting
Predictive modeling
Open Source Software
Time Series Analysis
Portfolio optimization
Statistical model
Strategy backtesting
Model risk analysis
View more
Data Science Consultant / Quantitative Researcher
Hull Tactical
2025-03-01-2026-07-01
  • Built and validated volatility forecasting models for E-mini S&P 500 (ES) futures, including GARCH a...
  • Built and validated volatility forecasting models for E-mini S&P 500 (ES) futures, including GARCH and Markov-switching GARCH models, supporting trading strategies tied to an estimated $100M–$250M in capital.
  • Engineered predictive features from alternative sentiment data, aligning sentiment signals with tick-level futures prices and applying sentiment-decay weighting to generate trading signals. Successfully advanced one model into live trading.
  • Developed market-microstructure pricing models using Databento tick data, analyzing buying and selling pressure to estimate fair value within the bid-ask spread.
  • Optimized a proprietary DEoptim-based algorithm through parameter tuning and data transformations, reducing execution time by 80% (from approximately 50 seconds to 10 seconds) and accelerating quantitative research workflows.
  • Performed out-of-sample validation, robustness testing, and model evaluation to assess predictive performance and ensure models demonstrated reliability before production deployment.
  • Designed rigorous quantitative experiments and signal validation frameworks to distinguish meaningful predictive signals from statistical noise, ensuring that only strategies demonstrating consistent performance on unseen data advanced toward live implementation.
Python
R
Machine Learning
View more
Python
R
Machine Learning
Statistics
Data Science
Forecasting
Data analytics
Performance Optimization
Feature engineering
Predictive modeling
Time Series Analysis
Statistical model
Optimization solvers
View more
Data Scientist / Quantitative Researcher
The Carlyle Group
2024-03-01-2024-12-01
  • Built sector-specific benchmarking models for U.S. investment teams, collaborating with data engineers to extrac...
  • Built sector-specific benchmarking models for U.S. investment teams, collaborating with data engineers to extract and analyze proprietary financial data from Snowflake.
  • Applied quantile-ranking and statistical analysis methods to identify outlier growth projections, helping investment teams challenge overly optimistic forecasts and improve investment assessments.
  • Designed a private equity deal-screening model to rank prospective investments against comparable companies using cash flow, operating expenses, and other key financial metrics.
  • Re-engineered and validated an analytics pipeline covering 50+ financial metrics, reducing processing time from days to hours and enabling overnight batch processing.
  • Developed outlier detection and treatment methodologies for private equity investment data, improving data quality and the reliability of downstream statistical analysis and financial modeling.
Python
SQL
R
View more
Python
SQL
R
Statistics
Data Science
Data analytics
Snowflake
Data Pipelines
Predictive Analytics
Statistical Analysis
Quantitative Analysis
Financial modeling
Data pipeline design
Statistical model
Ai/ml data pipeline qa
View more
PROJECTS
Volatility Forecasting Models for S&P 500 Futures
2025
The team needed better volatility forecasts and new signal sources for systematic E-mini S&P 500 strategies managing an estimated $100M–$...
The team needed better volatility forecasts and new signal sources for systematic E-mini S&P 500 strategies managing an estimated $100M–$250M. I tested GARCH and Markov-switching GARCH models to capture changing market regimes. Then I built a sentiment-signal pipeline that aligned alternative sentiment data with tick-level futures prices and applied decay weighting so older sentiment counted less. Every model went through out-of-sample validation before it was considered for trading, and one was promoted to live trading. Along the way I profiled and tuned a slow DEoptim-based optimization routine, cutting its runtime by 80% (from ~50s to ~10s) so research could iterate much faster.
R
Machine Learning
Statistics
View more
R
Machine Learning
Statistics
Data Science
Time Series Analysis
View more
AI-Powered Quantitative Research & Portfolio Optimization Platform
2018
Developed a quantitative research and portfolio optimization platform for systematic investment strategies, combining statistical modelin...
Developed a quantitative research and portfolio optimization platform for systematic investment strategies, combining statistical modeling, portfolio allocation, and historical performance evaluation. Built reusable Python and R components for portfolio backtesting, performance attribution, turnover analysis, risk measurement, and drawdown evaluation. Implemented optimization approaches including mean-variance optimization, risk parity, and momentum-based allocation strategies. The platform supports quantitative strategy evaluation through performance metrics such as CAGR, Sharpe ratio, volatility, maximum drawdown, and Calmar ratio, helping assess portfolio performance, risk exposure, and strategy robustness.
Python
R
NumPy
View more
Python
R
NumPy
Pandas
Machine Learning
Data Visualization
Time Series Analysis
Financial modeling
Portfolio optimization
Statistical model
View more