Hey there, I am

Victor Johnson

AI Engineer with 7+ years of experience and deep expertise in Generative AI, Retrieval-Augmented Generation (RAG), and multi-agent LLM systems. Proven track record designing scalable AI platforms for semantic search, document QA, and agentic workflows using Azure OpenAI, vector databases, LangChain, LlamaIndex, and LangGraph.
Experienced in LLM evaluation, grounding, guardrails, and low-latency production deployments using Python, FastAPI, and cloud MLOps.
Core: LLMs, RAG, Vector Search, Agentic Systems, LLM Evaluation, Python, Azure OpenAI, Spark, LangChain,LangGraph, MLOps.

England, United Kingdom

LinkedIn GitHub

Tech Stack

RAG & Knowledge Systems

Retrieval-Augmented Generation (RAG)
Semantic Search
Embeddings & Vector Search
Vector Databases
Hybrid Retrieval
Grounding & Guardrails
LLM Evaluation & Monitoring
RAGAS
DeepEval
LlamaIndex

Agentic AI & LLM Systems

Large Language Models (LLMs)
Generative AI
Prompt Engineering
Agentic Workflows
Multi-Agent Systems
Tool Use & Function Calling
LangChain
LangGraph
Chatbot Development
LLMOps
Azure OpenAI
Azure AI Foundry

Machine Learning & NLP

Machine Learning
Applied Machine Learning
Natural Language Processing (NLP)
Scikit-learn
Artificial Intelligence (AI)
Predictive Maintenance
PyTorch

Data Science & Analytics

Data Science
Python
SQL
Transact-SQL (T-SQL)
Statistical Modeling
Data Visualization
Financial Modeling

Data Engineering & Big Data

Apache Spark
Azure Databricks
Apache Kafka
Kafka Streams
Apache Airflow
Azure Data Factory
Elasticsearch
Azure SQL

Knowledge Graphs & Search

Knowledge Graphs
Knowledge Graph Data Engineering
Semantic Search & Information Retrieval

MLOps & Production AI

MLOps
MLflow
Model Monitoring
Model Drift & Bias Detection
Docker

Cloud & Platforms

Microsoft Azure
Azure Machine Learning
Oracle Cloud Infrastructure (AI Foundations)

Software & APIs

FastAPI
REST APIs
Software Development
Git
GitHub

Experience

Data Scientist (AI/LLM Focus)

Feb 2021 — Present

Company - SYMEUS LTD

England, United Kingdom · Hybrid | Industry - Finance

Architected and deployed end-to-end RAG systems using Azure OpenAI (LLMs & embeddings), Azure Vector Search, Azure SQL, LlamaIndex, LangChain, FastAPI, reducing manual research effort by ~50% and improving retrieval relevance by 25–35% while optimising latency to production-acceptable ranges. Delivered AI-driven semantic search and dynamic content ranking across international financial platforms serving 100K+ monthly users, enabling scalable, context-aware LLM-powered content experiences. Built agentic LLM workflows using LangGraph with tool orchestration, guardrails, and Monte Carlo simulations for probabilistic financial modelling and scenario analysis, contributing to ~20% revenue uplift. Applied statistical modelling and quantitative financial analysis techniques to support AI-driven decision systems and financial insight generation. Implemented automated LLM evaluation, regression testing, and drift monitoring (data, embeddings, retrieval) to detect hallucinations and silent quality degradation; reduced invalid responses by ~20–25%. Enhanced RAG retrieval by combining vector similarity with entity-aware ranking and structured relationship signals to improve contextual grounding and complex query handling. Developed scalable ingestion and preprocessing pipelines (chunking, tokenisation, embeddings, Mistral OCR) processing tens of thousands of structured and unstructured documents per run. Built and maintained Spark/Azure Synapse pipelines supporting ML/AI workloads and high-volume data processing, ensuring reliable training and inference pipelines.

Data Scientist (ML Focus)

Dec 2018 — Dec 2020

Company - Alstom

Bengaluru, India · On-site | Industry - Railways

Built and productionised machine learning models for predictive maintenance, including Remaining Useful Life (RUL), degradation modelling, time-series forecasting, and survival analysis for critical train components. Analysed millions of telemetry events per day to detect anomalies, sensor drift, and early failure signals, delivering ~5–10% maintenance cost savings. Extended component usable life by ~10–15% through improved failure prediction and optimised maintenance scheduling strategies. Applied Monte Carlo simulations to model failure uncertainty and maintenance scenarios, supporting data-driven preventive maintenance planning. Developed scalable data preprocessing and feature engineering pipelines and integrated ML model outputs into operational decision-support systems used by engineering and maintenance teams. Collaborated with cross-functional engineering and operations teams to deploy and monitor ML models in production environments.

Junior Data Scientist

Dec 2017 — Dec 2018

Company - Alstom

Bengaluru, India · On-site | Industry - Railways

Built a search system using Elasticsearch and Python (Flask), improving document retrieval efficiency by ~20%. Developed automated ETL pipelines using Apache Airflow and contributed to Spark and Kafka streaming workflows for high-frequency telemetry data. Created RPA workflows to automate SAP-based data processing, reducing manual effort and improving throughput.

Data Science Intern

Sep 2017 — Nov 2017

Company - Pi Revolutions

Bengaluru, India · On-site | Industry - Retail Tech

Supported data analysis and automation workflows for NFC-enabled billing kiosks Performed exploratory analysis to support process improvements

Intern

Aug 2016 — Sep 2016

Company - Alstom

Bengaluru, India · On-site | Industry - Railways

Automated routine business processes using VBA in Excel, ensuring compliance with internal data standards and improving efficiency, which saved more than 10 hours of reporting work each week