Data Science Application Developer

Sarath Lella

+1-571-***-**** ***********.****@*****.*** sarath-lella sarathlella sarathlella.portfolio

PROFESSIONAL SUMMARY

Results-driven AI/ML Engineer with 6+ years of experience in software development, data science, and machine learning. Adept at leveraging advanced ML and Generative AI techniques to deliver impactful business solutions, particularly in financial and telecommunications domains. Built a strong foundation in .NET and transitioned into AI and data science through hands-on experience with real-world healthcare and financial datasets. Demonstrated success at leading institutions like BNY Mellon and Charter Communications.

6+ years of professional experience developing scalable Python applications, leveraging key ML libraries such as NumPy, Pandas, Scikit-learn, TensorFlow, and Keras for end-to-end model development and deployment.

Experienced in prompt engineering techniques (zero-shot, few-shot, instruction-tuned prompts) and evaluation strategies (BLEU, ROUGE, perplexity) to guide LLM behavior for production-grade use cases.

Fine-tuned large language models (LLMs) using PEFT/LoRA techniques on domain-specific datasets such as MedQuAD and PubMed to enable context-aware QA and summarization.

Architected and implemented Generative AI solutions to automate financial report summarization and client advisory assistance

Designed and developed robust, scalable, and secure enterprise applications using .NET Core / .NET 6+, C#, and ASP.NET MVC / Web API, focusing on modular architecture, performance optimization, and maintainability across large codebases.

Proficient in building ML models using TensorFlow, PyTorch, and scikit-learn with deployment in Dockerized environments - Specialized in deployment, inference, tuning, and performance measurement of ML models across production environments.

Strong hands-on experience with MLOps tools like MLflow, SageMaker, and DVC for model lifecycle management

Familiar with Snowflake data pipelines, including Streams, Tasks, UDFs, and the Cortex platform for deploying ML models at scale.

Developed internal tools for ML explainability and model diagnostics using SHAP, LIME, and ELI5

Built scalable and production-ready APIs using FastAPI, Flask, and REST principles to serve models in real time

Solid foundation in cloud platforms (AWS, Azure) with experience deploying ML solutions using SageMaker and Azure DevOps

Created insightful dashboards using Power BI and Seaborn to communicate model findings to stakeholders and leadership teams

Built autonomous Agentic AI workflows using LangChain agents and tool integrations for multi-step reasoning, dynamic API calls, and decision-making automation.

Strong statistical foundation with hands-on experience applying hypothesis testing, time series analysis, and probabilistic modeling across regulated financial and healthcare datasets.

Designed and deployed cloud-native ML applications using AWS services like SageMaker, Bedrock, S3, Lambda, AWS Batch, Step Functions, SNS, SQS, EventBridge, and CloudFormation templates (CFT).

Built and maintained infrastructure as code (IaC) using AWS CloudFormation, Terraform, and OpenTofu for scalable cloud resource management.

Automated build, test, and deployment processes using Jenkins for continuous integration and delivery (CI/CD).

Hands-on experience using Git for source control, branching strategies, and collaborative development in Agile teams.

Expertise in infrastructure as code (IaC) services, including AWS CloudFormation and tools like Terraform or OpenTofu, for managing and provisioning cloud resources

Built, tested, and deployed containerized applications using Docker, optimizing for highly scalable distributed systems based on open-source technologies.

TECHNICAL SKILLS

Category

Skills

Programming Languages

Python, C#, JavaScript, SQL

ML/DL Frameworks

TensorFlow, PyTorch, Keras, Scikit-learn

Generative AI

LLMs, Transformers, Diffusion Models, GPT-4, RAG Systems

Cloud & DevOps

AWS (SageMaker, EC2, S3, RDS), Azure DevOps, GitHub Actions, Docker

Tools & Libraries

Pandas, NumPy, OpenCV, HuggingFace, MLflow, Airflow, Visual Studio

Databases

PostgreSQL, SQL Server, MongoDB, Oracle RDS

Visualization

Power BI, Matplotlib, Seaborn

Web & API

Flask, FastAPI, RESTful APIs

Version Control

Git, GitHub, GitLab

PROFESSIONAL EXPERIENCE

BNY Mellon, New York NY

Sr. AI/ML Engineer

May 2023 – Present

Designed and implemented a Generative AI-based intelligent document summarization system tailored for regulatory reports and investment memos by leveraging large language models (LLMs) and retrieval-augmented generation (RAG) pipelines, significantly reducing manual compliance review time and enabling financial analysts to gain faster insights, while adhering to the strict auditability, interpretability, and security requirements expected in the financial sector.

Engineered multi-agent systems using LangChain to automate legal document processing, integrating LLMs with external tools for RAG-based search, summarization, and report drafting.

Applied prompt engineering strategies for dynamic user intent recognition, retrieval reranking, and answer formatting in chatbot and search applications.

Developed explainable AI models (XAI) using SHAP and LIME to meet regulatory compliance requirements for transparency in financial decision-making.

Led development of ML pipelines for fraud detection and risk scoring, significantly improving model accuracy through ensemble learning

Integrated NLP models and transformer-based architectures to classify, summarize, and extract key information from internal documents

Conducted LLM fine-tuning and evaluation for GPT-based models to meet task-specific performance benchmarks, using datasets like MedQuAD and evaluating with BLEU and ROUGE.

Implemented model versioning and rollback strategies to ensure production model stability and compliance with audit and regulatory guidelines (e.g., SOX, Basel, GDPR).

Implemented an LLM-powered chatbot with RAG, enhancing internal response efficiency by over 60%

Developed scalable and automated ML workflows using Airflow, MLflow, and Docker, reducing model retraining time by 40%

Applied feature selection techniques and advanced hyperparameter tuning to improve prediction models across asset classes

Built predictive models to optimize loan default risk, policy lapse prediction, loss ratio modeling, mortality and morbidity risk, and customer segmentation.

Applied statistical analysis and ML models to optimize business operations and detect anomalies, fraud, and customer churn patterns

Collaborated with cross-functional teams to align AI systems with business KPIs, compliance, and audit-readiness requirements

Developed asynchronous inference APIs using FastAPI with Uvicorn and Gunicorn to support low-latency, high-availability deployments in financial environments.

Monitored model drift and deployed continuous evaluation pipelines using MLflow and SageMaker

Deployed secure and optimized REST APIs for ML microservices with FastAPI and OAuth2 authentication

Experienced in scaling ML pipelines and real-time inference systems to handle large volumes of data and predictions.

Developed and deployed ML services in AWS, leveraging S3 for data storage, SageMaker for model training/inference, Step Functions for orchestration, Lambda for event-driven processes, and Athena/Redshift for big data analytics.

Built and exposed RESTful APIs and microservices integrated with Azure App Services, AWS API Gateway, and OAuth2-based authentication (OpenID Connect), supporting millions of client requests with high availability and low latency.

Charter Communications, Stamford, CT

Data Analyst/System Developer

Aug 2022 – May 2023

Developed and deployed machine learning pipelines that processed multi-terabyte-scale customer network usage logs in near real-time, applying time-series forecasting and anomaly detection models to proactively identify service degradation and preempt outages, resulting in a 25% improvement in network reliability metrics and a major reduction in customer churn, critical to maintaining service-level agreements in a high-availability telecom environment.

Designed predictive models to forecast network load, detect anomalies, and optimize service uptime using time-series data

Developed end-to-end data science solutions including data wrangling, model development, and deployment via Azure ML

Built customer segmentation and churn prediction models using XGBoost and Random Forests integrated with Power BI for visualization

Leveraged SQL and Python-based pipelines for data ingestion and preprocessing of streaming customer usage data

Enabled A/B testing and uplift modeling to guide product development and network investment strategies

Used .NET Core and Python to build microservices integrating ML outputs with front-end dashboards and internal tools

Applied advanced techniques like ensemble learning (Random Forests, XGBoost) and deep learning (DNNs, RNNs, CNNs for time series forecasting) for improving model accuracy and robustness.

Designed asynchronous Python APIs to serve time-series anomaly detection models, supporting concurrent requests with optimized inference latency.

Created robust ETL workflows in Azure Data Factory and automated model refresh jobs.

Contributed to internal research on generative modeling techniques for synthetic log generation and rare event simulation, driving innovation in predictive maintenance.

Conducted feature importance analysis and implemented explainable AI techniques to promote stakeholder trust

Delivered presentations to leadership explaining ML model outcomes, insights, and actionables with data storytelling

Engineered real-time fraud detection systems using Kafka Streams, AWS Kinesis, and serverless Lambda functions to monitor transactions for anomalies.

Hansa Solutions, Hyderabad, India

.NET Developer & Data Analyst

Dec 2018 – Aug 2021

Led the design of an ML-powered decision support system for a Insurance services client that integrated logs, diagnostic codes, and historical sales performance to generate dynamic pathway suggestions and cross-sell opportunities using classification algorithms, Bayesian optimization, and clustering techniques—bridging the gap between Insurance intelligence and business strategy in a data-scarce environment with heavy compliance constraints.

Transitioned from traditional .NET development into AI/ML by building predictive and classification models on client sales and health data

Designed ML systems adhering to financial sector regulations, ensuring data privacy, model explainability, audit trails, and risk modeling guidelines (e.g., OCC guidelines, IFRS 17, Solvency II).

Containerized ML models and APIs using Docker and deployed them to AWS ECS/Fargate, Azure Kubernetes Service (AKS), and Google Kubernetes Engine (GKE).

Led statistical modeling efforts for pricing strategy optimization, utilizing A/B testing, uplift modeling, and regression-based elasticity estimation in compliance-heavy environments.

Deployed ETL pipelines that supported training of low-resource classification models fine-tuned on customer intent datasets using HuggingFace transformers.

Developed .NET MVC web apps that incorporated embedded ML inference engines for real-time recommendations

Designed anomaly detection systems to monitor transaction spikes and flag unusual business activity

Used Python for building ETL scripts and pandas-based analytics on structured and semi-structured data

Collaborated with product owners, actuaries, underwriters, risk officers, and compliance teams in Agile SCRUM environments to iteratively deliver impactful ML solutions in production.

Created dashboards in Power BI to support operational and strategic decisions informed by ML model results

Applied machine learning to improve client sales forecast accuracy using regression and time-series models

Collaborated with data scientists to deploy recommendation engines and customer insights modules

Integrated rule-based and ML-driven decision engines into existing enterprise reporting pipelines

Conducted multivariate statistical analyses to support product development and pricing strategies

Improved report delivery systems using SSRS and integrated visual ML model outputs

Led model retraining and validation workflows for evolving customer behavior models

EDUCATION

Master of Science in Data Science

University of Missouri-Kansas City, 2022

Bachelor of Technology in Computer Science

GMR Institute of Technology, Rajam IN, 2019

PROJECTS & OPEN-SOURCE

Heart Disease Risk Predictor – Deployed SHAP-explainable model with 81% ROC AUC

Early Diabetes Detection – Wearable-style feature dataset with >90% accuracy

Breast Cancer Classifier – Patch-level IDC detection using ensemble of CNNs (EfficientNet, ResNet, DenseNet) with test-time augmentation, explainability overlays (GradCAM), and >94% accuracy.

Medical Chatbot using LLMs – Fine-tuned transformer models on MedQuAD and PubMedQA, applied prompt engineering with LangChain, and deployed with RAG for multi-turn QA over custom knowledge base.

Clinical NER – Fine-tuned BioBERT for identifying medical entities in discharge summaries, integrating output into downstream pipelines for EHR processing.

References Available Upon Request

Contact this candidate