# Pulkit Saxena

> ML Engineer & Data Scientist with 5+ years of experience across fintech, telecom, and research. Expert in building production ML systems, LLM/Agentic AI pipelines, and MLOps infrastructure. $50K/year cost savings delivered, 500K+ users served, MS Data Science (RIT, 3.87 GPA), IEEE-published.

## About

Pulkit Saxena is a Data Scientist and ML Engineer currently at the University of Rochester, where he builds production data systems, anomaly detection pipelines, and LLM-powered workflows. He specializes in bridging experimental AI research and production-grade engineering.

- **Current role**: Data Scientist at University of Rochester (Aug 2024–Present)
- **Location**: Rochester, NY (open to remote/hybrid)
- **Contact**: https://linkedin.com/in/pulkitsaxena14
- **LinkedIn**: https://linkedin.com/in/pulkitsaxena14
- **GitHub**: https://github.com/pulkitsaxena14

## Professional Experience

- **University of Rochester** (Aug 2024–Present): Data Scientist. Built research analytics platform processing 800GB+ data. Implemented hybrid rule-based + Isolation Forest anomaly detection saving $50K/year. Deployed LLM pipelines improving data prep by 30%. Built GitLab MLOps Python Client with MLflow integration. Dockerized Dash dashboards on Posit Connect. Configured dbt for transformation workflows.
- **T-Mobile USA** (May–Aug 2023): Credit Risk Management Intern. Built automated Risk Strategy Simulation Tool (+40% productivity). RCA-inclusive alerts for policy deviations (−75% investigation time). Migrated legacy SAS pipeline to Python for MLOps. Analyzed 35% policy deviation using Snowflake, Databricks, Pandas.
- **RedCarpetUp (YC S15)** (Apr 2019–Jan 2021): Data Scientist. Led ML platform for 500K+ users, 40% MoM credit disbursal growth. XGBoost credit risk models with A/B testing across 4,000+ customers. Docker + Flask production deployment. EDA on 5TB customer data identifying ₹5.2M revenue opportunities.
- **RedCarpetUp (YC S15)** (Oct 2017–Mar 2019): Software Engineer. Built Flask/PostgreSQL CRM for 500K+ users. KYC image processing automating 80% of verification. Led team of 4–5 engineers through sprint planning and delivery.
- **Foodpost** (Jan–Mar 2016): Full Stack Developer Intern. RESTful APIs with PHP/Codeigniter.

## Technical Skills

- **AI & ML**: LangChain, Agentic AI, scikit-learn, MLflow, XGBoost, Isolation Forest, A/B Testing
- **Data Engineering**: dbt, Airflow, Dagster, ETL/ELT, SQLAlchemy, Alembic, Spark, PySpark, Schema Evolution
- **Cloud & Big Data**: GCP/BigQuery, AWS SageMaker, AWS S3, Snowflake, Databricks, PostgreSQL
- **Infrastructure & Tools**: Docker, DevContainers, Git, GitLab CI/CD, Flask, Dash, Posit Connect, Tableau, Linux
- **Languages**: Python, SQL, R, C++, Bash

## Education

- **MS Data Science** — Rochester Institute of Technology (2022–2024), GPA 3.87/4.0. Coursework: Machine Learning, Statistical Analysis, Big Data Analytics, Database Systems.
- **BE Computer Science** — Visvesvaraya Technological University (2012–2017). Coursework: Data Structures, Algorithms, Database Management, Software Engineering.

## Publications

- "Peer-to-peer Knowledge Sharing Platform for Farmers with Auto-Recommendation Feature" — IEEE SmartTechCon 2017. DOI: 10.1109/SmartTechCon.2017.8358498. URL: https://doi.org/10.1109/SmartTechCon.2017.8358498

## Projects

- **Vendor Invoice Anomaly Detection**: Isolation Forest + rule-based pipeline at UofR. $50K/year savings. Stack: Python, SQLAlchemy, Alembic, MLflow, Docker.
- **Research Analytics Platform**: 800GB+ BigQuery pipeline at UofR. +20% research media coverage. Stack: BigQuery, dbt, Dash, Tableau, LLMs.
- **Risk Strategy Simulation Tool**: T-Mobile. +40% productivity, −75% investigation time. Stack: Python, Pandas, Snowflake, Databricks.
- **XGBoost Credit Risk Models**: RedCarpetUp. 500K+ users, 40% MoM growth, ₹5.2M revenue identified. Stack: XGBoost, Docker, Flask, PostgreSQL, Airflow.
- **Expansive Insights — ChatGPT Analysis**: Analyzed 107,000 developer interactions with GenAI. Identified 16 single-turn + 8 multi-turn interaction patterns.
- **Mini-GPT**: Transformer built from scratch in PyTorch. Character-level language modeling on Shakespeare corpus.
- **Enterprise CRM System**: Flask/PostgreSQL CRM for 500K+ users. KYC image processing, audit trail schemas.

## Pages

- [About](https://pulkitsaxena14.github.io/) — Profile, skills overview, key achievements
- [Experience](https://pulkitsaxena14.github.io/resume.html) — Full resume with all work history, education, publications
- [Projects](https://pulkitsaxena14.github.io/projects.html) — ML projects portfolio with impact metrics
- [Resume PDF](https://pulkitsaxena14.github.io/resume.pdf) — Downloadable PDF resume