Vetted Retrieval-Augmented Generation Professionals

“Built and deployed a production LLM-driven Dungeons & Dragons game where the model acts as a dungeon master, adding a structured combat system and a macro-state tree to ensure campaigns converge to a clear ending. Fine-tuned Gemini 2.5 Flash on Vertex AI and deployed on GCP with Kubernetes, using RAG over DnD rules/spells plus multi-agent orchestration (intent-based routing between narrative and combat agents) to reduce hallucinations and improve reliability.”

“Built and deployed a production LLM/RAG clinical decision support system that enables real-time semantic search over unstructured EHR notes and delivers patient risk insights. Strong in healthcare-grade MLOps and compliance (HIPAA, PHI handling, encryption, RBAC, audit logs) and scaled embedding/retrieval pipelines using Spark/Databricks and Airflow. Partnered with clinicians via Power BI dashboards and explainability, contributing to an 18% reduction in patient readmissions.”

“Built and deployed an LLM-powered RAG document intelligence/search platform for banking risk & compliance teams, emphasizing sensitive-data handling, traceability, and conservative fallback logic to minimize hallucinations; deployed via Docker/REST on AWS and cut manual review effort by 35%. Also partnered with TD Bank marketing to deliver an AI customer segmentation solution that improved targeted campaign engagement by 18%.”

“Built an AI journaling app at HackCU 2025 featuring a speaking AI avatar with long-term memory via RAG (ChromaDB) and low-latency microservices coordinated through Kafka, including deployment under AMD/non-CUDA constraints using a quantized Llama 8B model. Also has Goldman Sachs experience deploying a Trade UI on Kubernetes with CI/CD rollback automation, plus a healthcare AI internship at CU Anschutz collaborating closely with physicians on diagnostic reasoning and dataset annotation.”

“Built and deployed an enterprise GenAI knowledge assistant over thousands of internal PDFs/reports using a RAG stack (GPT-4 + Hugging Face embeddings + vector DB) to reduce manual search and SME escalations. Uses LangGraph/LangChain to orchestrate modular agent workflows with relevance filtering and fallback handling, and applies rigorous evaluation (golden datasets, edge cases, A/B tests) with production monitoring metrics.”

“AI/LLM engineer with production experience at Charles Schwab building a RAG-based assistant to help 5,000+ reps answer complex financial policy questions. Implemented a multi-layer anti-hallucination approach (GNN-driven ontology/graph retrieval + citation-only answers) and compliance-focused guardrails (Azure AI Content Safety) in partnership with audit/compliance stakeholders.”

“Interned at Zonda where he built an AI-driven semantic search solution over ~280M housing/builder records. Iterated from local LLMs via llama.cpp quantization to a vector-embedding retrieval system, then boosted semantic accuracy with a custom spaCy NER layer and re-ranking, optimizing for latency through precomputation. Collaborated with economics-focused stakeholders to reduce manual document/paperwork time by enabling natural-language search over internal data.”

“ML/AI engineer with Capital One experience building production-grade customer segmentation and fraud detection systems combining NLP (transformers) and anomaly detection. Strong MLOps and orchestration background (PySpark ETL, MLflow, Airflow, Docker/Kubernetes, Azure ML) with real-time monitoring/alerting and performance optimizations like quantization and caching, plus proven ability to deliver business-facing insights through Power BI/Tableau for marketing stakeholders.”

“Built and productionized an LLM-powered PDF document Q&A system to eliminate manual searching through long documents, focusing on scalability and answer reliability. Implemented semantic chunking (using headings/paragraphs/tables), overlap, and preprocessing/quality checks to reduce hallucinations, and orchestrated the end-to-end pipeline with Airflow using retries, alerts, and parallel tasks.”

“Built and deployed a production LLM-powered internal documentation assistant using embeddings, a vector database, and a RAG pipeline to reduce time spent searching PDFs/manuals. Experienced in orchestrating end-to-end LLM workflows with Airflow/LangChain, improving reliability via monitoring/error handling, and driving measurable quality through retrieval and hallucination-focused evaluation metrics.”

“New grad who has shipped a production vision-language recommendation feature for a pet camera/mobile app, including building a tagged video dataset with human annotators and optimizing inference by FPS downsampling under device compute limits. Also built a multimodal MLLM benchmark using an LLM-as-judge (GPT-5-thinking) with a feedback loop, validated against human scoring, and measured post-feedback quality gains (12% average score improvement).”

“Built and deployed an LLM + RAG + memory system on a Furhat social robot, adding continuous face/voice recognition embeddings over WebSockets to enable persistent, natural conversations across sessions. Experienced working around real-world hardware/latency constraints and uses Datadog plus structured debugging/rollback practices for stabilizing customer-facing LLM workflows.”

“AI/ML engineer with experience at Accenture and Morgan Stanley, building production LLM systems (GPT-3 summarization) and finance-focused ML models (credit risk and trading anomaly detection). Combines MLOps depth (Docker/Kubernetes, AWS SageMaker/Glue/Lambda, MLflow, A/B testing, drift monitoring) with practical domain adaptation techniques like few-shot prompting and RAG/knowledge-base integration.”

“GenAI Engineer at Infosys who built and deployed a production multi-agent RAG system for a top-tier bank, scaling to ~50,000 queries/day with 99.9% uptime. Drove measurable gains (45% accuracy improvement, 30% API cost reduction) through open-source LLM fine-tuning, Pinecone indexing/retrieval optimization, and AWS-based MLOps/monitoring, and has experience enabling adoption via developer workshops and customer-facing collaboration.”

“Researcher from the University of South Dakota who built a production medical RAG system to help interpret model predictions by retrieving relevant clinical notes and medical literature, overcoming retrieval accuracy and imaging-dataset challenges through semantic chunking and metadata-driven indexing. Also has hands-on orchestration experience with Airflow and Azure Data Factory, plus a pragmatic approach to LLM evaluation and stakeholder-driven iteration.”

“Healthcare/clinical ML practitioner who built and productionized ClinicalBERT-based pipelines to extract and standardize oncology EHR data, improving downstream model F1 from 0.81 to 0.92 while controlling training cost via LoRA/QLoRA. Experienced orchestrating real-time AWS ETL/ML workflows (Glue, Lambda, SageMaker) and partnering with clinicians using SHAP-based interpretability, contributing to an 18% reduction in readmissions and full adoption.”

“McKinsey data/ML practitioner who led production deployment of an entity resolution + semantic search platform for unstructured finance and healthcare data, integrating with legacy systems under HIPAA constraints. Deep hands-on stack across transformers (spaCy/HF BERT), embeddings + FAISS, and production MLOps/workflow tooling (Airflow, Docker, CI/CD, Prometheus/Grafana), with reported gains of +30% decision speed and +25% search relevance.”

“Backend engineer who built and owned an AI-powered SMS survey platform for a nonprofit serving at-risk communities (internet-limited users), using Cloudflare Workers + Twilio and a state-machine survey engine. Scaled it to ~10k active users with near-zero downtime, added English/Spanish support, and iteratively improved LLM behavior (Claude 3.7 Sonnet) to handle nuanced, real-world SMS responses reliably.”

“AWS Solutions Architect intern who advised customers on securing a multi-tenant LLM-based SaaS, including isolation strategy tradeoffs and production guardrails against prompt injection. Has experience investigating a prompt-injection incident using logs/traces and TTP-style documentation, and designing scalable SDK/agent integrations via asynchronous worker architecture with prompt versioning.”

“AI/software engineer with experience at Visa building a real-time transaction fraud/risk scoring microservice in the card authorization path (Python, Kafka, Kubernetes on AWS) with strict 120–150ms latency constraints and reason-code outputs for downstream decisioning. Owns ML backend end-to-end (data/feature engineering, model training, deployment) and has demonstrated production reliability work including latency spike mitigation, SLO-based observability, drift monitoring, and safe fallbacks to rule-based decisions.”

“Built and owned "Cortex," an AI agent that helps users understand large GitHub repositories by mapping architecture and relationships between files/folders in minutes. Implemented an agentic, multi-stage prompt decomposition approach and validated it across open-source repos, while also doing legacy service modernization work involving dependency upgrades and refactors.”

“Siemens AI engineer who shipped production multi-agent LLM systems across cybersecurity and sustainability, including a vulnerability automation agent that cut manual work 70%. Deep in orchestration (LangGraph supervisor-worker state machines), reliability engineering (async fault tolerance, retries, spike handling), and rigorous evaluation (offline benchmarks, LLM-as-a-Judge improving label agreement 28.9%) with measurable production guardrails.”

“AI/ML engineer who has shipped production AI systems end-to-end, including an automated multi-channel (Gmail/WhatsApp/voice) candidate interviewing workflow and an enterprise RAG knowledge search platform. Demonstrates strong production rigor (monitoring, A/B tests, guardrails, schema validation, shadow testing) with quantified impact: ~60–70% reduction in interview evaluation time and ~20–30% relevance gains in RAG retrieval.”

“GenAI/LLM engineer who architected and deployed a production RAG “research assistant” for JPMorgan Chase’s regulatory compliance team, focused on safety-critical behavior (mandatory citations, refusal when evidence is missing). Deep hands-on experience with LlamaIndex, Pinecone, Hugging Face embeddings, LangGraph agent workflows, and metric-driven evaluation (golden sets, TruLens), including a reported 28% relevancy lift via cross-encoder re-ranking.”