I build production-oriented AI applications, machine learning systems, developer tools, and cloud-native services.
I am an AI Engineer based in Bengaluru, India, with experience building machine learning applications, generative AI systems, backend services, and full-stack products.
My work focuses on the intersection of:
- Machine learning and deep learning
- Generative AI, RAG, and AI agents
- MLOps and production model deployment
- Backend and cloud-native engineering
- Developer automation and intelligent tooling
I enjoy taking AI ideas beyond experimentation and turning them into reliable, usable systems.
- Building production-ready machine learning and generative AI applications
- Designing RAG pipelines, AI agents, and tool-using workflows
- Deploying models through APIs, containers, and cloud infrastructure
- Learning advanced MLOps, Kubernetes, distributed ML systems, and GPU programming
- Improving model monitoring, evaluation, retraining, and deployment workflows
An automated code-review system that analyses pull requests, retrieves related Jira or Azure Boards context, and generates both summary-level and file-level feedback.
Highlights
- GitHub Actions and GitHub App integrations
- Azure OpenAI-powered code analysis
- Jira and Azure Boards work-item retrieval
- Inline review comments and pull-request summaries
- Context-aware responses to mentions and review discussions
A multimodal interview assessment platform that evaluates candidate responses using audio, video, natural language processing, and computer vision.
Highlights
- Speech transcription and filler-word analysis
- Response relevance scoring
- Eye-contact, posture, smile, and confidence assessment
- React and Next.js frontend
- Python-based machine learning services
- Dockerized deployment architecture
A machine-learning pipeline for identifying anomalous operating conditions in automotive and equipment sensor data.
Highlights
- Autoencoder-based feature extraction and anomaly scoring
- XGBoost classification
- Time-series and sensor-data preprocessing
- Separate modelling strategies for different equipment types
- Interactive visualisation and CSV inference interface
A tool-using AI agent for reading and creating Jira issues from structured and unstructured input.
Highlights
- LangGraph and ReAct-based workflows
- Epic, user-story, and subtask creation
- Excel-driven bulk issue generation
- Human-in-the-loop confirmation
- Issue updates, comments, assignment, and linking
A cloud data architecture for ingesting, transforming, storing, and serving data for machine learning inference.
Highlights
- Azure Data Lake Storage
- Databricks and Delta Lake
- Medallion architecture
- Streaming and batch transformation pipelines
- Model training and inference workflows
- Machine learning and deep learning
- Natural language processing
- Computer vision
- Time-series anomaly detection
- Retrieval-augmented generation
- AI agents and tool calling
- Embeddings and semantic search
- Model evaluation and inference
- REST API design
- Authentication and authorization
- Microservices
- Asynchronous processing
- AI model serving
- Third-party API integrations
- Docker and Kubernetes
- CI/CD pipelines
- Azure App Service, Functions, VMs, Storage, and networking
- AWS data engineering services
- Model deployment and monitoring
- Databricks, Delta Lake, and data pipelines
- Linux server administration
I am interested in opportunities and collaborations involving:
- Machine learning engineering
- MLOps and ML platforms
- Generative AI and RAG
- AI agents and developer tooling
- Computer vision and NLP
- Backend engineering for AI systems
- GPU programming and ML systems research
- Portfolio: sonythomas.me
- LinkedIn: linkedin.com/in/njansony
- GitHub: github.com/STSonyThomas
- X: x.com/sony_national