Venkatalakshmi's Portfolio

🌙

Hello Everyone!

I'm
Venkatalakshmi Kottapalli

I leverage AI, ML, Data Scienc and software solutions to creatively
solve real-world problems. Join me in exploring innovative ideas and building impactful solutions!

About Me

AI/ML & Data Science Practitioner

Hi, I’m Venkatalakshmi Kottapalli. I build intelligent AI and ML systems, including RAG pipelines, LLM-powered agents, LLM fine-tuning, and scalable ML models, turning insights from data into impactful solutions.
🔹 Across of my roles, I’ve delivered 95% model accuracy, 40% ROI uplift, and contributed to 5 research papers.
🔹 I specialize in Machine Learning, Generative AI, Computer Vision, and Cloud Solutions, turning complex data into actionable insights and impactful AI-driven products.
🔹 Let’s connect to explore ideas, collaborations, or innovative projects!

More About Me

Technical Skills

Python SQL R Cypher

95%

PostgreSQL Neo4j FAISS

85%

Scikit-learn PyTorch TensorFlow Keras OpenCV

90%

LangChain LangGraph NLTK SpaCy

90%

Azure AWS GCP

85%

Power BI Visual Studio Code Jupyter

85%

Streamlit FastAPI Flask OOP Techniques

85%

MLOps MLflow CI/CD, GitHub Docker & Kubernetes

85%

Classification Regression Clustering Time-series Forecasting

90%

LLMs Prompt Engineering RAG Agentic AI Transformers

85%

Object Detection Image segmentation Image Generation

85%

Professional Skills

90%

Creativity

95%

Communication

80%

Problem-Solving

90%

Agile Framework

90%

Innovation

80%

Team Work and Leadership

View My exprience / Acedamics

My Research

PEFT LLM Fine-tuning with DEKF Large Language Models (LLMs)

This research explores fine-tuning transformer-based large language models using Decoupled Extended Kalman Filters (DEKF) for adaptive uncertainty estimation. Implemented in JAX, NNX, and Dynamax on Google Cloud TPUs. Achieved 42% lower computing, 38% reduced memory, and 1.8× faster convergence compared to LoRA.

📄 Set to Publish in February 2026

DEKF Dynamax NNX Jax GCP

Generating Images Using GAN and Diffusion Models

This research explores generating realistic images using GAN and diffusion techniques. It compares different models to assess image quality and generation efficiency.

📄 View Research Paper

GANs Diffusion Models Computer Vision Deep Learning Image Generation

Detection of Canine Cardiomegaly Using Deep CNNs

This research presents a CNN-based approach to detect enlarged hearts in canine chest X-rays, helping improve early diagnosis accuracy in veterinary care.

📄 View Research Paper

CNN Deep Learning Veterinary Imaging Cardiology AI in Healthcare

UNet-Based Model for Segmenting Bird Sound Spectrograms

This research proposes a UNet model to segment bird sound spectrograms, improving acoustic feature isolation for better species identification.

📄 View Research Paper

UNet Deep Learning Spectrogram Analysis Bioacoustics Segmentation

Detection of Key Points Localization in Medical Images

This research explores deep learning methods to detect key points in canine chest X-rays, automating vertebral heart score calculation to aid veterinary diagnosis.

📄 View Research Paper

Deep Learning Computer Vision Medical Imaging Veterinary AI

View All Publications

MyProjects

🤖 Multi-Agent Conversational AI System

Developed a multi-agent conversational AI platform integrating GPT-4, Retrieval-Augmented Generation (RAG) with ChromaDB, and CRM features to handle real estate, CRM, and general queries with personalized, context-aware responses.

GPT-4 ChromaDB FastAPI SQLite

Source Code

🤖 GenAI Agent

Developed an AI agent leveraging LLMs, LangChain, and LanGraph to automate order processing, account management, and inventory tracking, reducing manual workload by 60%.

LangChain LangGraph LLMs

Source Code

📚🔍RAG-Based Chatbot

Created a Retrieval-Augmented Generation (RAG) Chatbot using LLM and FAISS vector database, delivering 90% accurate, citation-backed responses sourced from domain-specific data.

LangChain FAISS LLM Prompt Engineering

Source Code

📘 Fraud Detection

Built an end-to-end ML pipeline using Python and SQL for financial fraud detection, applying supervised and unsupervised learning techniques to achieve 96% accuracy.

Python Scikit-learn Supervised models

Source Code

🔄 Migrating Postgres Database to Neo4j

Executed migration of relational data from PostgreSQL to Neo4j graph database, optimizing query performance and enabling complex relationship analysis using the Chinook dataset.

SQL Cypher Neo4j PostgreSQL

Source Code

🩺 SympCheck: Disease Prediction Chatbot

Designed an NLP chatbot using Hugging Face models for predicting potential diseases from reported symptoms, deployed with Azure and Streamlit for easy access.

Python Hugging Face Azure Streamlit

Source Code

❤️ Heart Disease Prediction

Developed a heart disease risk prediction model with 94% accuracy using Random Forest, featuring a web app and API using Streamlit for real-time assessment.

Python Random Forest Streamlit FlaskAPI

Source Code

📉 Customer Churn Prediction

Built a model to predict customer churn using customer behavioral data, integrated with FlaskAPIs and Streamlit UI to help improve retention strategies.

Python XGBoost FlaskAPI

Source Code

👥 Customer Segmentation

Performed customer segmentation using clustering techniques to enable targeted marketing and personalized customer experiences.

Python K-Means Clustering EDA

Source Code

🚗 Adoption Rates of EV Vehicles

Analyzed electric vehicle adoption trends and factors affecting growth for for 50 USA states using WebScraping, data science and visualization techniques.

Data Analysis BeautifulSoup Selenium

Source Code

🏠 House Price Prediction

Developed regression models using Scikit-learn to predict house prices from key features, achieving 92% accuracy after preprocessing and feature selection.

Feature Engineering Regression Seaborn

Source Code

🌊 Microplastic Detection in Water Sources

Built an ML model to detect microplastics in water sources, supporting environmental monitoring and research.

Python SQL Data Acquisition ETL

Source Code

📊 Impacts of Sleep on Mental Health – Data Analysis

Conducted data analysis on the relationship between sleep patterns and mental health using the NHANES dataset. Identified correlations between sleep duration and lifestyle factors to uncover key behavioral insights.

R EDA Data Analysis Seaborn Matplotlib

Source Code

Visit My GitHub

Experience

AI/ML Engineer

Peblink, New York, USA

Sep 2025 - Present

🔹

Implemented AI-driven optimization models for Goldman Sachs’ $500M Dallas campus initiative, identifying redundant management layers and improving operational efficiency.

🔹

Built ensemble ranking system (XGBoost, Random Forest) to prioritize relocation candidates, reducing analysis time by 40%.

🔹

Developed GPT-4 LangChain agents for relocation and retention scoring, automating 30+ workflows, reducing time by 25%.

🔹

Created interactive dashboards, heatmaps, and executive reports to visualize relocation scoring and cost-benefit analyses.

Visit Website

Machine Learning Co-op

ZSAnalytics LLC

May 2024 - Aug 2025

🔹

Designed and deployed an end-to-end machine learning lifecycle for email marketing using MLOps best practices, achieving 95% accuracy, 40% ROI improvement, and 98% customer retention.

🔹

Built ML pipelines for data ingestion, data preprocessing, feature engineering, training, evaluation, deployment, and automated retraining with Azure ML Studio and Azure Kubernetes Service (AKS).

🔹

Developed a retrieval-augmented generation (RAG) pipeline using LangChain, GPT-4, and vector database ChromaDB to index 100K+ documents, achieving Recall@K = 0.82 and reducing query latency from 6.2 s to 2.1 s.

🔹

Processed and analyzed large-scale datasets using SQL, NumPy, and Pandas, ensuring clean and reliable data for downstream modeling.

🔹

Performed exploratory data analysis (EDA) and visualizations with Matplotlib, uncovering hidden customer behavior patterns and campaign insights.

🔹

Applied feature engineering, creating 10+ new features that boosted model precision by 12%, enabling more accurate customer targeting.

🔹

Trained predictive models using linear algorithms and ensemble methods, evaluated with precision, recall, F1-score, ROC-AUC, and confusion matrix to ensure fairness across segments.

🔹

Tracked and compared model performance using MLflow for experiment tracking and model selection.

🔹

Implemented MLOps pipelines with Azure DevOps CI/CD, REST API integration, and Azure Monitor for automated deployment and troubleshooting.

🔹

Delivered insights through dashboards in Power BI, Excel, and PowerPoint, empowering data-driven marketing strategies.

Visit Website

Data Scientist

Cutso

Jan 2019 - Dec 2021

🔹

Collected, managed, and processed sales, customer, and inventory data using SQL and Python, ensuring accuracy and consistency for analytics and reporting.

🔹

Built ETL pipelines with SQL, Pandas, and NumPy to extract, clean, and structure large datasets for analysis.

🔹

Engineered preprocessing workflows that resolved 95% of missing values/outliers and corrected skewness, improving model performance.

🔹

Conducted exploratory data analysis (EDA) (univariate, bivariate, multivariate) to uncover customer behavior patterns, demand cycles, and a 10% seasonal variation in sales that informed demand planning.

🔹

Applied statistical techniques (Chi-square, ANOVA) to identify three key business drivers impacting revenue performance.

🔹

Built clustering models to segment customers into five groups, increasing customer retention by 18%.

🔹

Designed and delivered interactive dashboards with Matplotlib, Seaborn, and Power BI, improving reporting efficiency by 15% and enabling KPI tracking for business decisions.

🔹

Delivered actionable insights on pricing, promotions, and inventory optimization, driving measurable improvements in profitability and operational efficiency.

Visit Website

Education

Yeshiva University

Master’s in Artificial Intelligence

Jan 2024 - Dec 2025

🔹

GPA: 4.0 | Relevant Courses: Machine Learning, AI, Data Science, Neural Networks & Deep Learning, Advanced NLP, Cloud Computing with AWS.

🔹

Achievement: Certificate of Excellence in Machine Learning - Awarded for an innovative end to end fraud detection project.

🔹

Machine Learning: Built end-to-end supervised and unsupervised models using classification, regression, and clustering (Logistic Regression, Linear Regression, SVM, Random Forest, XGBoost, Gradient Boosting).

🔹

Generative AI / NLP: Built end-to-end GenAI systems including sentiment analysis, recommendation systems, similarity matching, and AI automation using RAG, agents, and multi-agent workflows (NLTK, LangChain, LangGraph, Prompt Engineering).

🔹

Computer Vision: Built end-to-end computer vision models for image classification, segmentation, and image generation using CNNs, GANs, and Transformers (PyTorch, TensorFlow, Keras, JAX).

Visit Website

Adikavi Nannaya University

Bachelor’s in Mathematics

June 2015 – May 2018

🔹

GPA: 3.8 | Relevant Courses: Statistics, Probability, Linear Algebra, Matrices, Calculus, Geometry.

🔹

Achievements: University First Rank - Secured the top position in the second year of undergraduate studies. 1st Prize - Project on Digital Transactions - Awarded first place at GIET College, competing against 100 teams from various institutions.

🔹

Mathematics & Analytical Skills: Completed a comprehensive program with a strong focus on calculus, linear algebra, probability, and statistics. Built a foundation in analytical thinking, problem-solving, and quantitative reasoning.

🔹

Data Structures & Algorithms: Gained practical experience designing optimized and efficient solutions, bridging theoretical mathematics with real-world applications.

🔹

Information Technology & Soft Skills: Developed a solid base in IT including HTML, and honed communication and teamwork skills through projects and presentations.

Visit Website