Generative Ai Senior

Christopher Smith Sylacauga, AL ***** • 713-***-****

**********************@*****.*** • https://www.linkedin.com/in/chris-s-88a852377 Senior AI/ML Engineer Expert in Generative AI, NLP & LLMs Cloud-Native AI Solutions (AWS

& Azure) AI Governance & MLOps Leader Driving Scalable, Impactful AI in Healthcare & Retail Top Skills

Generative AI & LLMs: GPT-3/4, LLaMA, T5, BART, RoBERTa, BioBERT, BioGPT, Prompt Engineering, Transfer Learning, Few-shot Learning, Zero-shot Learning, Self-supervised Learning, Adapter Tuning

AI Agents & Frameworks: LangGraph, LangChain, Open AI Agents SDK based multi-agents systems, Agent memory, context retention, function calling, and autonomous tool use, Modular plug-and-play pipelines for custome use cases

NLP & Chatbot Systems: Retrieval – Augmented Generation (RAG) using ChromaDB, FAISS, Pinecone, Chatbot systems that support PDF, DOCX and website ingestion, Tokenization (WordPiece, Byte-Pair Encoding), Embeddings (word2vec, GloVe, FastText, ELMo), Transformer Architectures, Multi- head Attention, Cross-Attention, Seq2Seq Models, TF-IDF, CountVectorizer, Named Entity Recognition, Sentiment Analysis, Azure Bot Service, Rasa, LangChain, OpenAI GPT APIs, Intent Recognition, Entity Extraction, Context Management, Multi-turn Dialogue Handling

Vision & Image Generation: Quantization and model converstion, Fine-tuning of Stable Diffusion, SDXL and Flux, Custom pipelines using ControlNEet and ComfyUI for stylized outputs

Model Training & Tuning: Fine-tuning LLMs(QWEN, LLaMA) on client-specific datasets, PEFT- based methods like LoRA for efficient adaptation, Support for evaluation pipelines and alignment tuning

Voice & Multimodal Interfaces: Real-time voice interfaces using Whisper, Azure Speech and WebSockets, Text-to-speech (TTS) with ElevenLabs,

Machine Learning Algorithms: XGBoost, LightGBM, CatBoost, Random Forests, SVM, Neural Networks (CNN, LSTM, GRU), GANs, Diffusion Models, PCA, Fourier Transforms, Feature Engineering, Hyperparameter Tuning (Optuna, Bayesian Optimization, GridSearchCV)

AI Engineering & MLOps: Model Lifecycle Management, MLflow, Automated Monitoring, Model Drift Detection, Bias Detection & Mitigation, Explainable AI (SHAP, LIME, Integrated Gradients), CI/CD Pipelines (Jenkins, GitLab CI/CD, Azure DevOps), Containerization (Docker, Kubernetes, AKS, EKS)

Vector Databases & Semantic Search: Pinecone, Weaviate, FAISS, Vector Similarity Search, Embedding Retrieval, Semantic Search, Large-scale Unstructured Data Indexing

Programming Languages & Frameworks: Python (3.x), TensorFlow, PyTorch, Scikit-learn, Pandas, NumPy, FastAPI, Flask, Gunicorn, uWSGI, ONNX

Communication & Collaboration: Cross-functional Team Collaboration, Stakeholder Management, Agile Methodologies, Product Ownership Alignment, Compliance Coordination (HIPAA, CCPA), AI Governance

Cloud Platforms & Services: AWS (SageMaker, Lambda, Glue, Redshift, EKS, Fargate), Azure (Azure ML, Databricks, Cognitive Services, AKS, Azure Monitor), Terraform, CloudFormation Work Experience

Instacart, Remote

Senior AI Engineer (Contract) 07/2024 - Present

Architected and deployed scalable AI/ML solutions for retail platform migration on AWS Cloud, leveraging AWS SageMaker, Lambda, and Glue to build resilient ETL pipelines and model workflows, accelerating feature rollouts by 35% and improving system reliability by 25%.

Designed, fine-tuned, and productionized state-of-the-art large language models (LLMs) including GPT-3/4, T5, BART, RoBERTa, and domain-adapted BioBERT, using advanced transfer learning, prompt engineering, few-shot/zero-shot learning, self-supervised learning, and adapter tuning to enhance NLP-driven personalization and support systems.

Developed NLP pipelines with tokenization methods (WordPiece, Byte-Pair Encoding), embedding techniques (word2vec, GloVe, FastText, ELMo), and transformer-based architectures (Multi-head Attention, Cross-Attention) to drive sophisticated language understanding and generation.

Engineered conversational AI chatbots with Azure Bot Service, Rasa, and LangChain, integrating OpenAI GPT APIs for generative dialogue, context management, intent recognition, entity extraction, and multi-turn conversation handling—improving customer support automation and reducing response time by 30%.

Designed and implemented vector similarity search and retrieval systems using vector databases such as Pinecone, Weaviate, and FAISS, enabling efficient semantic search over embeddings generated by LLMs and improving retrieval accuracy and latency for large-scale unstructured data.

Led AI governance initiatives by implementing model risk assessment, bias detection and mitigation, data privacy compliance (HIPAA, CCPA), and ethical AI frameworks in collaboration with cross-functional teams; ensured AI models met company and regulatory standards throughout the development lifecycle.

Collaborated closely with product owners, data scientists, compliance officers, and engineering teams to define AI strategy, validate model outputs, prioritize feature development, and align AI capabilities with business goals and ethical guidelines.

Applied advanced training techniques including mixed precision training, gradient checkpointing, and distributed training on AWS SageMaker and EKS, scaling transformer-based models efficiently and reducing compute costs without sacrificing accuracy.

Integrated explainability and transparency tools such as SHAP, LIME, Integrated Gradients, and attention visualization into AI workflows, enabling stakeholders to interpret model decisions confidently and satisfy audit requirements.

HCA Healthcare, Nashville, TN

Lead AI Engineer 11/2022 – 06/2024

Led a high-impact AI engineering team of 7, managing full project lifecycles and client engagements, driving AI innovation to transform healthcare operations through data-driven insights and generative AI solutions.

Architected and deployed scalable, HIPAA-compliant AI/ML solutions on Microsoft Azure, utilizing Azure Machine Learning, Azure Databricks, and Azure Kubernetes Service (AKS); integrated Azure Cognitive Services with advanced custom models for NLP, vision, and speech applications.

Developed state-of-the-art generative AI platforms, including fine-tuning and deploying large language models (LLMs) such as GPT-3/4, LLaMA, T5, and BioGPT tailored for clinical data synthesis, automated report generation, and conversational agents, leveraging transfer learning, prompt engineering, and zero/few-shot learning techniques.

Built advanced generative adversarial networks (GANs) and diffusion models for medical image synthesis, anomaly detection, and data augmentation to improve downstream diagnostic model performance and reduce data scarcity issues.

Engineered custom transformer architectures and optimized attention mechanisms for multimodal healthcare data fusion, combining electronic health records (EHR), imaging, and clinical notes to enhance predictive analytics and patient outcome forecasting.

Pioneered conversational AI applications with Azure Bot Service, integrating LangChain frameworks and OpenAI GPT APIs to build intelligent chatbots for patient engagement, virtual health assistants, and clinical workflow automation, significantly improving user interaction quality and efficiency.

Designed and implemented vector search solutions using vector databases like Pinecone and Weaviate to enable efficient semantic retrieval over clinical text embeddings, enhancing information discovery and decision support for healthcare practitioners.

Implemented end-to-end MLOps pipelines with Azure ML Pipelines for generative AI model lifecycle management, including automated fine-tuning, bias detection, hallucination mitigation, and continuous evaluation; monitored model drift and data quality using Azure Monitor and custom alerting.

Promoted explainability in generative AI outputs through novel XAI techniques like integrated gradients, saliency maps, and counterfactual reasoning, ensuring trustworthiness and regulatory compliance in sensitive healthcare environments.

Fostered an innovative culture by integrating open-source generative AI toolkits, contributing to 15+ active GitHub repositories, and streamlining collaborative workflows with CI/CD automation via Azure DevOps, improving team productivity by 40%.

Morgan Stanley, New York, NY

Machine Learning Engineer 05/2018 – 10/2022

Engineered and optimized predictive models using XGBoost, LightGBM, CatBoost, Random Forests, SVM, and Neural Networks on large-scale datasets accessed via PostgreSQL and AWS Redshift; utilized advanced SQL queries, window functions, and stored procedures to perform efficient feature extraction and aggregation on transactional and time-series financial data.

Designed and automated complex feature engineering pipelines incorporating rolling windows, lag features, Fourier transforms, PCA, feature crossing, and target encoding using Pandas, NumPy, and SQL; orchestrated ETL workflows with Apache Airflow and AWS Glue, ensuring consistent data ingestion from PostgreSQL and cloud data lakes.

Conducted comprehensive hyperparameter tuning using Optuna, Bayesian optimization, and GridSearchCV; validated models rigorously via time-series split, k-fold cross-validation, and stratified sampling to mitigate overfitting and data leakage in production-ready ML models.

Built and fine-tuned NLP pipelines for document classification, sentiment analysis, and entity recognition using TF-IDF, CountVectorizer, and transformer models (BERT, DistilBERT) with Hugging Face Transformers; trained and deployed models via AWS SageMaker leveraging PostgreSQL for metadata and annotation storage.

Evaluated model performance using metrics such as ROC-AUC, Precision-Recall, F1-score, Matthews correlation coefficient, and log loss; applied calibration techniques (Platt scaling, isotonic regression) and threshold optimization to improve predictive accuracy and reduce false positives on business-critical models.

Integrated explainability frameworks including SHAP and LIME for feature importance and decision boundary visualization, aiding compliance teams in model validation and regulatory reporting.

Maintained reproducibility and governance using MLflow experiment tracking and model registry backed by PostgreSQL; implemented automated data quality checks with Great Expectations and monitored model health with AWS CloudWatch and custom SQL-based alerting.

Developed RESTful model serving APIs with FastAPI, containerized with Docker, and deployed on AWS SageMaker Endpoints and AWS Fargate; collaborated with DevOps to implement CI/CD pipelines using Jenkins, GitLab CI/CD, and infrastructure provisioning with Terraform and AWS CloudFormation.

Reddit, San Francisco, CA

AI Engineer 11/2016 – 04/2018

Developed large-scale ML models for content ranking and subreddit recommendation using Python

(3.6), Scikit-learn, XGBoost, and LightGBM, optimizing for engagement and click-through rates; performed feature extraction on user behavior logs using Apache Spark (PySpark), Pandas, and NumPy, processing data from Redshift, Hive, and S3.

Built and automated data processing pipelines using Apache Airflow, SQLAlchemy, and Jinja templating, handling daily ingestion, transformation, model training, and batch scoring; managed dataset versioning and lineage through structured S3 folder conventions, timestamped artifacts, and custom experiment logs.

Trained NLP models for comment classification and subreddit moderation using TensorFlow 1.x, Keras, and PyTorch 0.3, implementing LSTM, GRU, CNN, and early attention-based architectures; used word2vec, GloVe, fastText, TF-IDF, CountVectorizer, and custom tokenization for input representation.

Deployed trained models via Dockerized REST APIs using Flask, Gunicorn, and uWSGI, exposing inference endpoints to internal consumers; optimized real-time inference performance using ONNX, NumPy broadcasting, batch prefetching, joblib, and shared memory arrays, with production deployment to AWS ECS and EC2 Auto Scaling Groups.

Integrated real-time prediction services with Kafka and Redis-based data streams, using gRPC, Protobuf, and Thrift protocols for low-latency communication with upstream feed-ranking services; implemented health checks, timeout handling, and circuit breakers to meet service SLAs under variable load.

Evaluated and tuned models using A/B testing frameworks, offline metrics (precision, recall, ROC- AUC, log-loss), and threshold optimization; contributed to drift detection, feature monitoring, and retraining triggers based on model confidence scores, user feedback signals, and data freshness metrics. Education

Bachelor of Science in Computer Science, Graduation Year (2016), Columbia University, New York, NY

Contact this candidate