Shrey Verma
AI/ML Engineer
ABOUT ME
Meet Shrey Verma, the data scientist that wears his love for numbers on his sleeve and his passion for health in his heart. With a stack of research papers on the ethics of AI by his side, Shrey's mission is to connect the world with data and change the game forever.
But don't let his nerdy persona fool you - behind those glasses and lab coat, Shrey is a fierce problem-solver who never shies away from a challenge. After all, he lives by the wise words of Steve Jobs - "We are here to put a dent in the universe."
When he's not crunching numbers and making scientific breakthroughs, you can find Shrey exploring new trails and cuisines, always on the lookout for his next adventure. Whether he's hiking up a mountain or coding his way to the top, Shrey is always up for a challenge - and always has a witty quip up his sleeve.
SKILLS
Programming Languages: Python, JavaScript, SQL, Java, C++, Haskell, Apex
Databases & Processing: PostgreSQL, MySQL, MongoDB, Apache Spark, Redis, PL/SQL
Packages & Tools: Docker, Kubernetes, Helm, Terraform, AWS EKS, Flask, FastAPI, RESTful APIs
Web Dev: React.js, Next.js, Express.js, HTML, CSS, TypeScript, Linux, Bash, Blockchain
Cloud & DevOps: AWS (EKS, Lambda, S3, RDS), GCP (Cloud Run, BigQuery), CI/CD
EXPERIENCE
July 2023 -July 2024
Ernst & Young (EY)
Tech Consultant, Senior Analyst
-
Aided in fine-tuning EY’s internal LLM using OpenAI, integrating RAG with FAISS for real-time knowledge retrieval and incorporating user feedback loops, ensuring complete compliance on EY’s data security policies.
-
Implemented an XGBoost model which was Dockerized and then deployed via FastAPI for inventory optimization for wire manufacturer, reducing POs and transportation costs for the client by 3%.
-
Automated data pipelines using PL-SQL, and Oracle R12.2 for a bonded warehouse to enhance data capturing, tracking, and visualization, resulting in 3% savings in cash flow.
August 2022 -December 2022
Software Development Intern
ATMS Co and LLP
-
Created a financial index for 4,000 MSMEs data stored in AWS S3 using Linear Regression, K-means clusters, and Random Forest to predict revenue trends and classify financial stability.
-
Deployed the index with weighted metrics (e.g., revenue growth, profit margins), achieving 0.92 correlation with real world and 95% confidence, via MLflow on Google Cloud Platform (GCP) for MSME trend tracking.
April 2022 - July 2022
Data Analytics Intern
Data Sutram
-
Engineered data pipelines using Apache Kafka and Spark to integrate real-time satellite and alternative data, and developed Naive Bayes models to identify optimal locations for drugstore openings.
-
Deployed models via CI/CD pipelines with Jenkins on AWS EC2 and visualized data using SQL and Tableau, achieving an 18% sales boost in pinpointed stores.
June 2021 -April 2022
Data Engineer Intern
RD&X Network
-
Engineered data pipelines using Apache Kafka and Spark to integrate real-time satellite and alternative data, and developed Naive Bayes models to identify optimal locations for drugstore openings.
-
Deployed models via CI/CD pipelines with Jenkins on AWS EC2 and visualized data using SQL and Tableau, achieving an 18% sales boost in pinpointed stores.
Have partnered with
Projects
1
Influenza Vaccination Coverage | Pfizer AI Team
-
Built a full-stack dashboard with Express.js and React.js, deploying optimized Random Forest and CatBoost models for influenza vaccination analysis, achieving 94% R2, and refining Pfizer’s campaign strategies.
-
Conducted large-scale data preprocessing, feature engineering, and handled class imbalance leveraging SMOTE and ADASYN, enabling the identification of key demographic trends that led to a projected $150k in additional revenue.
2
Investment Recommendation Model | JPMorgan Wealth Management
-
Built and deployed a pairwise neural network with k-core extraction using Flask and AWS EKS for startup-investor recommendations, improving accuracy by leveraging funding type and investment history.
-
Delivered data-driven insights to improve investor relations and competitive positioning by achieving 93% accuracy and 98% precision, enhancing client satisfaction with tailored startup recommendations.
3
Hospital Mortality Forecasting
-
Developed and deployed a mortality prediction model based on patient-doctor interactions using MIMIC-III dataset, leveraging XGBoost and Random Forest, and deployed it using Azure Kubernetes Service (AKS).
-
Optimized model performance with 5-fold cross-validation, hyperparameter tuning, and class imbalance handling, achieving an ROC-AUC of 0.95 and an F1-score of 0.93.
4
Human Protein Atlas CV
-
Developed a CNN ensemble for protein localization utilizing NFNet L2 and a modified ResNet 200d with CBAM, achieving 0.62 mAP and 0.70 F1 score.
-
Designed and implemented a HIPAA compliant, production-grade, real-time federated learning framework deployed through AWS EKS.