cv

Machine Learning Engineer

• Led the development of verifiable LLM on Qwen 2.5 with no overhead of computation and designed Automated testing using Multi Context Protocol (MCP) sampling.
• Led the Quantization and post-training team and Infrastructure on Open-source LLM Models.
• Developed the first Verifiable Agentic AI with proof of inference, enabling verifiable model predictions to meet legal and insurance liability requirements.
• Engineered a comprehensive automated SR&ED auditing documentation pipeline using Google's Flan T5, seamlessly integrating GitHub, Slack messages, and Jira to streamline employee documentation.
• Designed and implemented a ranking system based on DeepFM and BERT, utilizing Reinforcement Learning with Human Feedback (RLHF) and Direct Preference Optimization (DPO), containerized with Docker.
• Developed Autonomous Index Agent with optimizing liquidity and slippage.
• Improved inference time by 23% by developing a Commit-Reveal strategy on Kubernetes using GCP instances.
• Developed Cosmos SDK modules for decentralized AI inference platforms.
• Developed a Retrieval-Augmented Generation (RAG) system using LLaMA 3 for the R&D team specializing in Zero-Knowledge Cryptography deployed on Graph VertexAI.
• Developed Verifiable Ray-tracing physical simulation based on Nvidia Warp, Google DeepMind Mujoco, and Autodesk XLB frameworks.
• Applied structured pruning and quantization techniques on time series prediction models using NVIDIA's pruning methods for 8-billion parameter LLaMA and ResNet models without loss of accuracy.

Machine Learning Developer

• Developed an Trading Bot, with enhanced risk mitigation strategies to prevent market crashes, leveraging FastDQN reinforcement learning to outperform human traders by 185%.
• Implemented multi-objective bi-strategy West America's rail-highway transportation optimization on high-performance computing cluster using Gurobi multi-threading and shell scripting focusing on class-based Supply Chain Management.
• Implemented a multi-objective bi-strategy optimization model using hierarchical reinforcement learning, improving conventional Multi-Armed Bandit problems by 13% with Meta's PEARL framework.
• Instructed 'Statistical Data Analysis in Healthcare' using Python (EHealth 705), with materials available at: github.com/shirin-shahabi/PythonTutorial_Ehealth705.

Data Scientist

• Developed a scalable, robust, and interpretable ensemble tree-based algorithm for clustering over 4 million customers by leveraging Spark for distributed transaction data processing, resulting in increasing MAU from 4M to 8.5M in under a year.
• Led cross-team collaboration with the Product Team to design an end-to-end Marketing Data Modeling and ETL Pipeline using Django and ORM Design.
• Defined +50 financial KPIs and identified company priorities and targeted measures for strategic growth.
• Utilized Web Engage for Customer Journey Mapping, achieving a 35% increase in conversion rate.
• Conducted market research and customer survey analysis, driving a 10% increase in user engagement.
• Collaborated on Power BI and EDA Course for Managers, promoting dashboard adoption for data-driven decision-making.
• Conducted Instagram and Twitter sentiment analysis, extracting key insights to inform marketing strategies, resulting in a 30% increase in retention rate.

Financial Analyst

• Designed a comprehensive business plan for a local +12 million dollar hotel construction project.
• Conducted cash flow projections and performed a feasibility study.

Data Analyst

• Contributed to a visionary product with over $100 million in funding for a fully automated shopping experience (Low-Cost Amazon Go), emphasizing indoor location tracking, in collaboration with Rocket Internet.
• Optimized inventory reordering for 15 hypermarkets with a fully automated algorithm for +8,000 SKUs.
• Collaborated on loyalty, fraud detection, retention, targeted marketing, and churn prediction programs.
• Mentored two interns in the Data Team, now successful Data Scientists one pursuing PhD in University of Michigan and the other doing Master of Computer Science at Waterloo University.
• Delivered +100 optimized SQL queries, benefiting Marketing, Operations, and Finance teams.
• Established real-time operational KPIs with automated email reports for timely insights, ensuring data accuracy.
• Defined packaging strategies and presented enhanced user experiences, resulting in a direct 10% increase in revenue.

M.Sc.

McMaster University

Computer Science

MBA

Sharif University

Finance

B.Sc.

Sharif University

Industrial Engineering