Looking for a job

Tristan Nagan

@tristannagan

Message

I build privacy-preserving ML, LLM, and document-intelligence systems for financial services.

South Africa

Message

What I'm looking for

I'm focused on building privacy-preserving AI, LLM, and document-intelligence systems for regulated financial-services environments, with reliable production deployment, structured extraction, and governance-aware workflows.

I'm building a privacy-preserving mortgage document-intelligence pipeline for Alpha7x through Stryv Growth, extracting and validating data from complex loan-origination documents.

I develop end-to-end workflows for segmentation, orientation handling, local OCR, classification, structured extraction, quality assurance, and borrower-level document association. I deploy open-weight multimodal models privately with vLLM and Kubernetes to keep PII-sensitive data out of external APIs.

At FNB, I designed and deployed machine-learning systems for fraud detection, behavioural analytics, and personalised digital-banking experiences. I built Spark-based pipelines in Scala and PySpark, supported conversational-AI and retrieval workflows, and operated ML workloads within governance, risk, and compliance requirements.

Earlier, I developed real-time and batch fraud-data pipelines using Spark Streaming, Kafka, and Java, sustaining 98%+ uptime and sub-300ms latency for production services. I also contribute to applied work in conversational AI, recommendation systems, and geospatial fraud detection.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning/AI Engineer

Current

Stryv Growth

May 2026 - Present (4 months)

Designing and developing a privacy-preserving mortgage document-intelligence pipeline for extracting, validating, and structuring data from complex loan-origination documents. Building end-to-end workflows for document segmentation, OCR, classification, field extraction, and quality assurance.

Python VLLM Kubernetes MongoDB OCR Document Classification Structured Extraction Open Weights Models Prompt Engineering JSON Outputs Validation

Data Scientist

First National Bank (FNB)

Mar 2023 - Apr 2026 (3 years 1 month)

Designed and deployed machine-learning systems for fraud detection, behavioural analytics, and personalised digital-banking experiences in a regulated financial-services environment. Built scalable Spark-based data pipelines and supported production LLM and conversational-AI initiatives.

Python Scala PySpark Spark DataRobot Kubernetes OpenShift Prompt Engineering RAG Human In The Loop Fraud Detection Behavioral Analytics

Data Engineer

First National Bank (FNB)

Jan 2021 - Feb 2023 (2 years 1 month)

Developed and maintained real-time and batch data pipelines for online and mobile banking fraud detection platforms. Implemented distributed streaming solutions using Spark Streaming, Kafka, and Java, achieving 98%+ uptime and sub-300ms latency.

Spark Streaming Kafka Java Scala Data Pipelines Fraud Detection Performance Testing Monitoring