AI Engineer - End-to-End LLMs, RAG, & MLOps Expert

Sowmya Sree

+1-940-***-**** ****************@*******.*** sowmyasree Sowmya-5G4

PROFESSIONAL SUMMARY

Machine Learning and AI Engineer with 4 years of experience delivering agentic LLM systems and production ML at scale. Built end-to-end RAG pipelines, tool- using agents, and multi-provider LLM flows (OpenAI/Anthropic/Azure) with CI/CD, monitoring, and governance across AWS, Azure, and GCP. Strong foundation in NLP, multimodal learning, and predictive analytics, with focus on AI governance, fairness, and secure deployments. EXPERIENCE

Machine Learning Engineer Jul 2024–Present

Bank of America Dallas, TX, USA

• Executed and productionized agentic LLM + RAG risk-stratification/propensity models with LangChain/LangGraph + XGBoost on Databricks (Spark) and features in Snowflake/BigQuery, improving clinical-style prioritization by 15% via Airflow retraining, eval harnesses, and gated promotion reviews.

• Configured and deployed agent orchestration + model-serving APIs with FastAPI on Kubernetes (HPA/autoscaling, canary/blue-green), sustaining 10K+ req/day <100ms; integrated MLflow traces/metrics, policy routing/fallback across OpenAI/Anthropic, and SLO/SLA guardrails.

• Structured retrieval + feature pipelines on Databricks (PySpark) over Delta/Parquet with Snowflake marts and vector indexes; added Great Expectations, chunking/embeddings, and DQ SLAs to improve training/inference quality across 50M+ records with compliant masking.

• Implemented CI/CD with Jenkins/GitHub Actions for Poetry/wheels, MLflow Model Registry promotions, Terraform/Helm env-specific configs, and prompt/tool versioning, reducing release errors by 40% and lifting F1 by 18% through gated validation and shadow testing.

• Spearheaded prediction and agent observability with Prometheus/Grafana + MLflow (latency, drift, stability, cost), adding SHAP explainability, audit trails, hallucination flags, and PII/PHI controls to align with Responsible/Explainable AI and enterprise risk standards. Machine Learning Engineer Jun 2022–Aug 2023

Infosys Chennai, India

• Devised and operationalized RAG chat/NLP pipelines using LangChain + Azure OpenAI and Hugging Face, packaged to Python wheels, embedded in FastAPI services, reducing manual review by 70% and standardizing handoffs with prompt libraries and tool registries.

• Launched transformer models on SageMaker and EKS with GPU autoscaling; added multi-LLM routing/fallback, validation suites, and registry workflows to cut infra cost by 20% while meeting throughput/latency SLAs and improving user-facing answer quality.

• Deployed dockerized microservices with Helm for AI inference APIs across staging and production, while conducting adversarial robustness evaluations on NLP/LLM pipelines to mitigate evasion and poisoning risks, accelerating release cycles by 35% with minimal downtime.

• Reconstructed legacy ML workflows into Docker/Helm microservices with Airflow/Databricks orchestration and Snowflake/BigQuery sinks, introducing IaC and rollout policies that accelerated delivery by 35% and hardened multi-tenant production standards. Data Scientist Jan 2021–Jun 2022

MAQ Software Remote, India

• Devised feature extraction strategies and applied statistical techniques on unstructured customer data with semantic indexing + embeddings, uncovering actionable patterns that improved decision-making efficiency by 20% and enabled searchable knowledge.

• preparation and delivery of 5+ ML prototypes blending cloud workflows, reproducible notebooks, and clear visual storytelling, reducing review cycles by 25% and aligning stakeholders via concise metrics, baselines, and acceptance tests. PROJECTS

Churn Prediction Engine for Subscription-Based Products

• Built an ETL training serving churn pipeline with Airflow + Databricks (Spark), XGBoost, MLflow Model Registry, and FastAPI/Kubernetes serving, improving retention by 15% with RAG-assisted explanations, drift monitors, and phased canary releases. NLP Feedback Insights Platform

• Architected a Transformer-powered RAG feedback insights platform with LangChain, Snowflake feature marts, vector search, and Prometheus/MLflow dashboards; packaged components as reusable Python libs and automated evals for promotion readiness. SKILLS

Programming & Scripting: Python, Java(Spring Boot), JavaScript, R, Bash, SQL, C, OOP (Object-Oriented Programming) Cloud & DevOps: AWS (SageMaker, EC2, S3, Lambda, EKS, CloudWatch), Azure (Azure ML, DevOps), GCP (BigQuery), Jenkins, Git, GitHub Actions, GitLab CI/CD, Terraform AI/ML Frameworks & Tools: Scikit-learn, TensorFlow, PyTorch, Keras, FastAPI, LangChain, RAG, vector DBs, Streamlit, MLflow, Docker, Kubernetes, Helm, Multimodal, LLM fine-tuning Data Engineering & Pipelines: Apache Spark, Snowflake, Databricks, Airflow, AWS Glue, Azure Data Factory, ETL Workflows Databases & Monitoring: PostgreSQL, MySQL, MongoDB, ElasticSearch, Prometheus, Grafana, Azure Monitor AI/ ML Techniques & Analysis: Classification, Regression, Clustering, Unsupervised Learning, Semi-supervised Learning, Autoencoders, Object Detection (YOLOv5/v8, Faster R-CNN), Image Segmentation, Feature Engineering, NLP (BERT, GPT, Transformers), Statistical Modeling, Hypothesis Testing, EDA, Visualization (Matplotlib, Seaborn, Plotly). Adversarial Attacks (poisoning, evasion), AI Fairness & Bias Mitigation, Secure Model Deployment, AI Auditing

& Logging, Data Privacy & Anonymization

Development Practices: Agile, Scrum, JIRA, REST APIs, Jupyter Notebooks, Linux, Security Principles, AI Security Compliance CERTIFICATIONS

Microsoft Certified: Azure Fundamentals (AZ-900) Azure Data Scientist Associate (DP-100) Azure AI Engineer Associate (AI-102) AWS Certified: Solutions Architect – Associate AWS Certified Machine Learning – Specialty Oracle Certified: OCI Generative AI Professional EDUCATION

Master of Science, Computer Science University of North Texas Bachelors, Computer Science Engineering JNTUK, India

Contact this candidate