Open to opportunities

Nicholas Keenan

@nkeenan38

Message

Senior AI/ML engineer building production LLM systems with low-latency, observable inference.

United States

Message

What I'm looking for

I’m looking to build scalable, observable LLM and ML platforms—end-to-end RAG, evaluation, and low-latency inference—where strong engineering, automation, and distributed systems design turn models into reliable real-world products.

I’m a Senior AI/ML engineer with 9+ years building production machine learning and LLM-powered systems across finance, insurance, and enterprise SaaS. I focus on end-to-end AI delivery—data pipelines, model training and fine-tuning, RAG architectures, and real-time inference services.

I specialize in deploying large-scale ML and conversational AI using Python and modern API infrastructure, with FastAPI, Django, and Kubernetes-backed microservices. I build evaluation and optimization loops that validate accuracy/latency trade-offs, including model conversion to ONNX Runtime to preserve reasoning and compliance behaviors.

In my recent work, I fine-tuned Llama 3.3 with LoRA on Vertex AI, automated RAG data processing with LlamaIndex and Pinecone embeddings, and instrumented production inference with OpenTelemetry for full distributed tracing. I also designed MCP-based agent integration layers that connect LLM agents to CRM and enterprise financial systems.

Across insurance and tax tech, I’ve shipped backend and distributed systems in Python, C#, TypeScript, Rust, and Go—building event-driven services, scalable workflows, and multilingual/ML-powered pipelines. I’ve driven measurable impact (e.g., +10% multilingual accuracy, 40% faster return processing, p99 under 200ms) and delivered strong engineering practices, including CI/CD automation and SOC 2 / ISO 27001 compliance.

Experience

Work history, roles, and key accomplishments

Current

Senior Software Engineer

Current

Gail

Oct 2023 - Present (2 years 8 months)

Fine-tuned Llama 3.3 with LoRA on Vertex AI and built end-to-end RAG pipelines (LlamaIndex + Pinecone) with agent tool-calling for financial Q&A and document intelligence. Deployed low-latency inference on GCE using Docker/FastAPI/Kubernetes with OpenTelemetry tracing and built an MCP integration layer plus a multi-channel voice AI pipeline.

LoRAs Retrieval Augmented Generation (RAG)LangGraph Vertex AI Pinecone fastAPI Docker Kubernetes OpenTelemetry

Senior Software Engineer

Lula Technologies, Inc.

Oct 2021 - Oct 2023 (2 years)

Built API-driven insurance workflows and a FastAPI-based driver-risk decisioning pipeline using scikit-learn and XGBoost with real-time inference on GKE/Pub-Sub. Improved multilingual email classification accuracy by 10% and added observability to keep p99 latency under 200ms with server error rates below 0.01%.

fastAPI scikit learn XGBoost Flan T5 GKE Google Cloud Pub Sub A B Testing .NET Core Terraform

Senior Software Engineer

Taxfyle

May 2019 - Oct 2021 (2 years 5 months)

Developed C#/.NET backend services and APIs and introduced ML-powered document automation using PyTorch and spaCy, accelerating return processing by 40% across the CPA network. Built DevSecOps infrastructure with Concourse CI, Terraform, and Kubernetes, reduced compute costs by 70% via ephemeral Go preview environments, and supported the company’s first SOC 2 Type II and ISO 27001 audits with zero

C# .NET PostgreSQL PyTorch SpaCy Concourse Terraform Kubernetes Go ISO 27001

Software Engineer Intern

Taxfyle

May 2017 - May 2019 (2 years)

Engineered cross-platform mobile applications for iOS and Android using C#/Xamarin and React Native, supporting Taxfyle’s growth to 100,000+ users and 200+ CPA firms. Built Python internal tools to automate operational workflows, and partnered on QA/testing for 30+ mobile app features across iOS and Android.