AI/ML Engineer LLMs, RAG, MLOps, Cloud Platforms

Harshitha Nellore

AI/ML Engineer

Overland Park, KS, USA +1-913-***-**** ******************@*****.*** LinkedIn Professional Summary:

AI/ML Engineer with 3+ years of experience building and deploying scalable machine learning solutions across AWS, Azure, and GCP. Strong expertise in Python, developing production-grade APIs using FastAPI, and managing deployments with Docker and Kubernetes (AKS/ACI/ECS). Experienced in building Agentic RAG platforms using LangChain, LlamaIndex, Hugging Face Transformers, and fine-tuning LLMs with LoRA/QLoRA, improving retrieval accuracy by 35%. Skilled in MLOps practices using MLflow/DVC for experiment tracking and model versioning. Also experienced in computer vision pipelines using OpenCV, YOLOv9, MediaPipe and predictive analytics using Prophet, XGBoost, LightGBM, delivering reliable solutions with monitoring via Azure Monitor.

Technical Skills:

• Programming & Scripting Languages: Python, C++, Java, Bash, JavaScript

• Machine Learning & Deep Learning Frameworks: PyTorch, TensorFlow, Scikit-learn, XGBoost, LightGBM, Prophet, Transformers, PEFT, LoRA, vLLM

• Web Development & APIs: FastAPI, RESTAPI, React, HTML, CSS

• Computer Vision & Multimedia: OpenCV, YOLO, MediaPipe

• Libraries: Pandas, NumPy, SpaCy

• AI & NLP Tools: LangChain, LlamaIndex, RAG, Prompt-Engineering, AI-Guardrails, CrewAI, XAgent, MCP

• MLOps & Model Management: MLflow, DVC, Docker, CI/CD, Vertex AI

• Databases & Data Storage: SQL, PostgreSQL, MongoDB, Athena, Big Query, Snowflake

• Automation Tools: Zapier, Workato, Make

• Cloud Platforms & Services: Azure (ML Studio, AKS, ACI), AWS (SageMaker, S3, Lambda), GCP (Vertex AI,BigQuery, Cloud Run)

• Embedding Models & Vector Databases: Pinecone, FAISS, Nvidia-NV-Embed Professional Experience:

Wells Fargo Jan 2025- Present

AI/ML Engineer Overland Park, kansas

Technologies: LLM, VLM, MCP, A2A, LangChain, RAG, Azure PostgreSQL(pgvector), Vertex AI, Codey, Kubernetes, Azure Blob Storage, Azure Monitor

• Designed and deployed LLM-powered Agentic RAG platform using LangChain / LlamaIndex, enabling secure knowledge retrieval workflows and improving response accuracy through hybrid semantic + keyword search.

• Fine-tuned domain-adapted language models using PEFT / LoRA / QLoRA with Hugging Face Transformers to increase precision and reduce hallucination for enterprise Q&A use case.

• Built scalable model inference services using Python + FastAPI with session-based memory and routing, delivering low-latency LLM responses for real-time user queries.

• Implemented vector retrieval pipelines using Azure PostgreSQL (pgvector), FAISS/Pinecone, and optimized embeddings (Nvidia NV-Embed) to strengthen contextual relevance and ranking quality.

• Developed MLOps CI/CD pipelines integrating Docker, MLflow, DVC, automated model versioning, artifact tracking, reproducible training, and secure release approval.

• Developed and optimized complex SQL queries to extract, transform, and validate financial and operational datasets; built interactive dashboards using Power BI and Excel (PivotTables, Power Query, advanced formulas) to support reporting, forecasting, and business decision-making

• Developed secure data ingestion and feature pipelines using SQL/PostgreSQL, BigQuery, Snowflake, Athena, enabling high-quality training data, auditability, and repeatable model evaluation.

• Integrated enterprise-grade security controls including RBAC/least-privilege access, secure token-based authentication, and restricted access to finance documents containing PII.

• Audit logging and traceability by capturing user query logs, retrieval sources, model versions, and response metadata for compliance and governance review.

Hexaware June 2021-Dec 2022

Junior ML Engineer Hyderabad, India

Technologies: Python, Prophet, Docker, Blockchain, OpenCV, XGBoost, Docker, ETL, CV Pipelines, YOLOv9, Pytorch, Lambda, RDS, Azure Analytics, ACI

• Worked on day-to-day Python development and support for data pipelines by cleaning, validating and standardizing mining/industrial logs before loading into reporting layers and dashboards.

• Built and maintained containerized services using Docker and supported deployments on AWS ECS / Azure ACI, including environment updates, config fixes, and handling runtime issues during releases.

• Automated data processing and validation workflows using Python and C#, improving reporting accuracy and reducing manual effort,supported software implementation and integration efforts by performing data mapping, testing, and post-deployment validation in enterprise environments.

• Collaborated with cross-functional finance and IT teams to configure and support enterprise reporting and planning systems, ensuring data consistency across consolidation, budgeting, and management reporting processes.

• Supported regular monitoring and incident handling by checking job failures, pipeline delays, abnormal spikes and performing quick fixes/retries; coordinated with team for RCA and preventive actions.

• Developed computer vision workflows using OpenCV and YOLOv9, including preprocessing, model confidence tuning and validation for object detection in industrial environments.

• Implemented safety compliance checks using MediaPipe pose estimation and integrated detection outputs into alerting/reporting modules for operational visibility. Projects: June 2023 – Aug 2023

Video-Based Golf Posture Detection & Tracking System (Computer Vision) Project Description:

Developed a video-based golf posture detection system to evaluate swing posture from recorded videos. Built a complete pipeline for frame processing, subject tracking, and pose/keypoint extraction to compute posture metrics. Ensured stable results during continuous motion using temporal smoothing and tracking consistency.

• Built an end-to-end golf swing video posture detection pipeline using Python and OpenCV, including frame extraction and preprocessing. Implemented subject tracking to maintain player consistency across frames during fast motion and occlusions.

• Used pose/keypoint detection to extract body landmarks and calculate posture metrics (spine angle, shoulder tilt, hip rotation, swing alignment). Applied temporal smoothing (EMA/Kalman-style) to reduce jitter and improve stability under motion blur, lighting changes, and camera shake. Education:

Master of Science – Computer Information Systems Jan 2023 – Dec 2024 University of Central Missouri, Missouri, USA

Certificates:

• Microsoft Certified: Azure Administrator Associate (AZ-104) o Hands-on experience managing cloud infrastructure, compute, storage, networking, and monitoring to support scalable and reliable AI/ML workloads.

• Data Science Using Python and R

o Practical training in data analysis, statistical modeling, machine learning, and visualization using Python and R for real world analytical and predictive use cases.

Contact this candidate