Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I'm Justin Pettit, a Principal AI Engineer based in Flint, Michigan, United States, with 10 years of experience designing and deploying scalable AI and ML systems. I specialize in agentic and LLM-powered architectures, model fine-tuning and alignment, multimodal document intelligence, and production-grade conversational AI, LLMOps and MLOps that deliver measurable business impact across enterprise and startup environments. I'm passionate about building robust, auditable AI solutions, mentoring teams, and driving responsible AI through red teaming, evaluation pipelines, and real-time monitoring. I've led end-to-end MLOps and LLMOps implementations using MLflow, SageMaker, Kubernetes, and cloud services, delivering faster deployments, lower costs, and higher production reliability, while collaborating across healthcare, finance, e-commerce, and education domains.…I'm Justin Pettit, a Principal AI Engineer based in Flint, Michigan, United States, with 10 years of experience designing and deploying scalable AI and ML systems. I specialize in agentic and LLM-powered architectures, model fine-tuning and alignment, multimodal document intelligence, and production-grade conversational AI, LLMOps and MLOps that deliver measurable business impact across enterprise and startup environments. I'm passionate about building robust, auditable AI solutions, mentoring teams, and driving responsible AI through red teaming, evaluation pipelines, and real-time monitoring. I've led end-to-end MLOps and LLMOps implementations using MLflow, SageMaker, Kubernetes, and cloud services, delivering faster deployments, lower costs, and higher production reliability, while collaborating across healthcare, finance, e-commerce, and education domains.

Justin Pettit

AI Engineer, Data Scientist, Full Stack Developer, +5





I'm Justin Pettit, a Principal AI Engineer based in Flint, Michigan, United States, with 10 years of experience designing and deploying scalable AI and ML systems. I specialize in agentic and LLM-powered architectures, model fine-tuning and alignment, multimodal document intelligence, and production-grade conversational AI, LLMOps and MLOps that deliver measurable business impact across enterprise and startup environments. I'm passionate about building robust, auditable AI solutions, mentoring teams, and driving responsible AI through red teaming, evaluation pipelines, and real-time monitoring. I've led end-to-end MLOps and LLMOps implementations using MLflow, SageMaker, Kubernetes, and cloud services, delivering faster deployments, lower costs, and higher production reliability, while collaborating across healthcare, finance, e-commerce, and education domains.…I'm Justin Pettit, a Principal AI Engineer based in Flint, Michigan, United States, with 10 years of experience designing and deploying scalable AI and ML systems. I specialize in agentic and LLM-powered architectures, model fine-tuning and alignment, multimodal document intelligence, and production-grade conversational AI, LLMOps and MLOps that deliver measurable business impact across enterprise and startup environments. I'm passionate about building robust, auditable AI solutions, mentoring teams, and driving responsible AI through red teaming, evaluation pipelines, and real-time monitoring. I've led end-to-end MLOps and LLMOps implementations using MLflow, SageMaker, Kubernetes, and cloud services, delivering faster deployments, lower costs, and higher production reliability, while collaborating across healthcare, finance, e-commerce, and education domains.

Available to hire

I’m Justin Pettit, a Principal AI Engineer based in Flint, Michigan, United States, with 10 years of experience designing and deploying scalable AI and ML systems. I specialize in agentic and LLM-powered architectures, model fine-tuning and alignment, multimodal document intelligence, and production-grade conversational AI, LLMOps and MLOps that deliver measurable business impact across enterprise and startup environments.

I’m passionate about building robust, auditable AI solutions, mentoring teams, and driving responsible AI through red teaming, evaluation pipelines, and real-time monitoring. I’ve led end-to-end MLOps and LLMOps implementations using MLflow, SageMaker, Kubernetes, and cloud services, delivering faster deployments, lower costs, and higher production reliability, while collaborating across healthcare, finance, e-commerce, and education domains.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Language

English

Fluent

German

Advanced

Japanese

Advanced

Work Experience

Founding AI Engineer at XUNA AI

April 1, 2024 - December 1, 2025

Architected and deployed XUNA Voice/SMS, a production-grade platform leveraging LiveKit, Deepgram STT, GPT-4 for reasoning, LangGraph for multi-agent orchestration, Pinecone with Cohere re-ranking for semantic retrieval, ElevenLabs TTS, and Redis on GCP Cloud Run to automate end-to-end business phone ops. Fine-tuned domain LLMs with PEFT + QLoRA on LLaMA 3.1 using Direct Preference Optimization (DPO) to improve task adherence and cost efficiency. Implemented an agentic AI architecture with LangChain/LangGraph, MCP for tool invocation, and inter-agent coordination via Google ADK. Built a document ingestion pipeline using Textract and Unstructured.io for layout-aware extraction, reducing manual processing by 60% and boosting automation accuracy by 33%. Created a Neo4j knowledge graph (50k+ entities) with a hybrid retrieval pipeline, cutting domain hallucinations by 35%. Established end-to-end ML pipelines with MLflow, DVC, and GitHub Actions enabling weekly retraining with automated eval

Principal AI Engineer at Opinov8

January 1, 2022 - February 1, 2025

Led 20+ agentic AI systems, designing multi-agent orchestration frameworks (LangChain, LangGraph, CrewAI, Haystack, Semantic Kernel) to automate workflows across healthcare, e-commerce, enterprise IT, EdTech, and finance. Built multimodal virtual support agents for B2B SaaS enabling real-time dialogue flow, user behavior adaptation, and task handoff across APIs and UIs using headless browser control and HID emulation. Implemented distributed fine-tuning pipelines for open-source LLMs (Llama, Gemma) using QLoRA and Instruction Tuning on SageMaker, accelerating convergence by 45% and reducing domain hallucinations by 40%. Developed red-teaming pipelines (DeepEval, Guardrails AI, LangSmith) for responsible AI in regulated domains. Built a document-understanding agent with multimodal parsing (LlamaParse, Landing AI) improving extraction accuracy across 1M+ financial records. Delivered agent-driven automation for e-commerce market research via LLM-based web crawlers, trend summarizers, and

Senior AI Engineer at Progress Software

January 1, 2018 - December 1, 2021

Developed real-time fraud detection platform for e-commerce using TensorFlow, scikit-learn ensembles and AWS SageMaker, achieving a 90% reduction in chargebacks and 80% fraud prediction accuracy for Fortune 500 retailers. Built an AI-driven customer service chatbot with Flask, Azure Cognitive Services and React, reducing average call center volume by 35%. Designed a hybrid credit decisioning engine combining regulatory rules with ML risk scoring via Flask microservices on Kubernetes, with MLflow for model versioning and Weights & Biases for monitoring. Implemented a product recommendation engine using TensorFlow embeddings with PostgreSQL and Redis caches, delivering ML-driven segmentation and merchandising. Engineered Python ETL/feature pipelines (Pandas, NumPy) processing 10M+ daily transactions for real-time fraud scoring. Trained and deployed BERT-based NLP models for financial document processing, boosting query understanding from 74% to 91%.

ML Engineer at Amazon

June 1, 2014 - December 1, 2017

Built large-scale recommendation models using DSSTNE and Apache Spark to personalize product suggestions for 100M+ users, contributing to substantial revenue. Developed XGBoost-based ranking models for Search, increasing top-3 CTR by 3.5% and boosting conversions. Trained multilingual ASR and NLU models to expand voice assistant into five international markets. Created real-time fraud detection with ensemble models, reducing false positives and preventing millions in losses. Engineered time-series forecasting for 1M+ SKUs using XGBoost/Prophet to optimize inventory and improve in-stock rates by 12%. Designed gradient-boosted classifiers to identify fake reviews, removing 100K+ fraudulent entries monthly.

Principal AI Engineer at Zazmic

January 1, 2022 - December 1, 2025

Led end-to-end development of 20+ agentic AI systems; designed multi-agent orchestration using LangChain, LangGraph, CrewAI, AutoGen, Google ADK, Semantic Kernel, Haystack and Azure cloud services to automate workflows across healthcare, e-commerce, enterprise IT, EdTech and finance. Implemented Reinforcement Learning with Verifiable Rewards (RLVR) to train models on structured reasoning tasks, improving reliability on SQL and code-generation workflows. Built and integrated 5+ document-understanding agents with multimodal parsing and reflection-augmented RAG (RefRAG/Hybrid-GraphRAG), enabling iterative retrieval refinement and evidence validation, increasing extraction accuracy on 1M+ financial records. Established cloud-native data governance using Microsoft Purview’s Atlas store, automated scanners, and Azure Data Factory lineage extraction with RBAC-based access controls for auditable metadata and compliance during large-scale migrations. Designed distributed fine-tuning pipelines

Founding AI Engineer at Digital Wellness Institute

June 1, 2020 - November 1, 2021

Designed and productionized a large-scale recommendation and ranking system using TensorFlow Wide & Deep models trained on user interactions and content metadata, deployed via AWS SageMaker endpoints; feature pipelines built on S3, Glue, and Airflow with low-latency inference served through ECS and Redis caching. Developed predictive models on healthcare datasets (EMR records, wearable sensors) to improve chronic illness risk prediction. Built an internal AI-powered finance assistant using Flask microservices, Azure Cognitive Services, and a React frontend for natural-language querying, document retrieval, and workflow assistance. Implemented a hybrid credit decisioning engine combining regulatory rules with ML risk scoring on Kubernetes, with MLflow for model versioning and Weights & Biases for performance monitoring. Created a product recommendation engine using TensorFlow embeddings with PostgreSQL and Redis caching, enabling ML-driven customer segmentation and merchandising. Develo

Data Scientist at Datadog

January 1, 2018 - June 1, 2020

Built real-time anomaly detection models using SARIMA and STL decomposition in Python (scikit-learn, NumPy, Pandas) on Apache Kafka and Cassandra clusters, reducing mean-time-to-detection by 40%. Productionized DBSCAN clustering and MAD-based outlier detection pipelines with PySpark on AWS EMR for 14,200+ customers. Developed DDSketch probabilistic percentile algorithm in Python, Go, and Java with <1% relative error on P99 latency, published at VLDB 2019 and integrated into production Kubernetes workloads. Implemented NLP-based log clustering using TF-IDF vectorization and tokenization in Apache Spark, reducing unstructured log noise by 95% and accelerating incident resolution for Fortune 500 clients. Deployed Holt-Winters and robust regression forecasting models via TensorFlow Serving on AWS EKS, enabling predictive capacity alerts across 350+ integrations and contributing to 146% net dollar retention. Designed graph-based root cause correlation engine using Neo4j and PostgreSQL, surf

AI Engineer at Amazon

August 1, 2015 - December 1, 2017

Built large-scale recommendation models using DSSTNE and Apache Spark to personalize product suggestions across 100M+ users, contributing to over 35% of e-commerce revenue through improved cross-sell and engagement. Developed XGBoost-based ranking models for Search to improve top-3 CTR by 3.5% and drive gains in customer conversion. Trained multilingual ASR and NLU models via transfer learning to expand voice assistant into five markets, achieving production-grade performance across English, German, and Japanese locales. Created a real-time fraud detection system using ensemble models on transactional and behavioral features, reducing false positives by 15% and preventing millions in chargebacks. Engineered time-series forecasting for 1M+ SKUs to optimize inventory and in-stock rates by 12% across fulfillment centers. Built gradient-boosted classifiers to identify fake reviews, removing 100K+ fraudulent entries monthly and restoring trust in ratings.