Prashobh Balasundaram
About Prashobh
Prashobh Balasundaram is a seasoned IT delivery leader with a proven track record of successfully executing complex, high-stakes IT projects across diverse domains. He is the founder of WordJog and brings extensive expertise in large-scale data engineering initiatives, GenAI-powered solutions tailored for banking and advanced data workflows, robust Big Data infrastructure deployments, and comprehensive cloud architecture designs.
His consistent history of on-time, on-budget deliveries underscores a deep capability in driving transformative technology outcomes for enterprise clients. Prashobh specializes in fraud detection systems (FICO Falcon) onboarding for financial institutions, cloud architecture leveraging Microsoft Azure, Google Cloud Platform, and Amazon Web Services, and cutting-edge GenAI solutions that enhance data engineering productivity.
A University of Edinburgh graduate with an MSc in High Performance Computing, Prashobh combines deep technical expertise with strong leadership skills. He has led teams at major organizations including Loblaws, Citibank, TD Bank, Deloitte, and Google, delivering innovative solutions that drive business value.
Recent Key Achievements
MerchOS - Multi-AI Agent Solution (Loblaws, 2025)
Led the design and delivery of MerchOS, a multi-AI agent solution for merchandising use cases that generates product insights and flyer proofing descriptions. Part of the AI Retail OS suite, this solution uses Gemini 2.5 Pro/Flash models via a LangGraph-based architecture with Langfuse traceability and evaluation layers. Designed Explainable AI agents with hallucination detection and Agent State Graph conditional routing.
NeuraRec Project - Regulatory Data Pipeline Modernization (Citibank, 2025)
Led strategic direction for modernizing regulatory data pipelines using AI/ML/GenAI to improve data quality and troubleshoot compliance workflows. This use case cuts developer effort for troubleshooting reconciliation errors by more than 40% using a graph refinement algorithm with LLMs as a Judge pattern. Successfully deployed to production with Hadoop/Spark/Impala and FastAPI-based design.
Cortex AI - Deloitte's Self-Service GenAI Platform (2024)
Built Cortex AI, Deloitte's self-service production-grade GenAI platform hosted on hybrid cloud AWS/OpenAI/GCP Bedrock. Implemented GenAI security standards (prompt shields, RBAC, input sanitization) and architected scalable, cost-optimized AI applications. Developed RFP Synthie, an AI tool automating RFP reviews with over 96% detection accuracy.
DagUI - GenAI-Powered Data Engineering Platform
Developed DagUI, a GenAI platform to enhance data engineering productivity using context graphs for lineage tracking, code generation, and feature change impact analysis. The platform accelerates data pipeline delivery by generating production-ready code using GenAI technology, enabling teams to ship faster with higher quality.
Technical Expertise
Generative AI & ML
- • LangGraph, LangChain
- • GPT-4, Claude, Gemini
- • MLOps, MLflow, W&B
- • PyTorch, Transformers
Data Engineering
- • Apache Spark, PySpark
- • Databricks, Airflow
- • Kafka, Beam, DataFlow
- • BigQuery, Hive, Impala
Cloud Platforms
- • AWS (Bedrock, SageMaker)
- • Google Cloud Platform
- • Microsoft Azure
- • Kubernetes, Docker
Notable Projects & Experience
Fraud Detection & Financial Systems
Implemented FICO Falcon fraud management systems for TD Bank and President's Choice Bank, including real-time Kafka integration, Spark/Cobrix for mainframe file parsing, and fraud scoring solutions. Developed cheque fraud detection using OCR and Keras, and designed data vault architectures for multi-hop fraud analytics.
Large-Scale Data Platform Leadership
Led teams at Cloudera managing 8 customer accounts worth $8M annually. Designed and deployed multi-petabyte enterprise data lakes for financial clients, including Manulife's Cloudera Data Platform on Azure. Optimized Twitter's log pipeline at Google, reducing data volume by 20-25% through compression.
High Performance Computing
Worked on IBM BlueGene supercomputing systems, managing BlueGene/L systems and LSF schedulers. Developed and optimized software for BlueGene/Q, published book chapters on performance analysis tools, and collaborated with leading universities on molecular dynamics simulations using OpenMP/MPI.
Education & Certifications
Education
- • MSc High Performance Computing - The University of Edinburgh, UK
- • BTech, Electronics - Cochin University of Science and Technology, India
Key Certifications
- • Databricks Certified Apache Spark Developer (Python)
- • Gen AI Certified Developer (Virtusa/Citibank)
- • Google Cloud Platform: Certified Professional Data Engineer
- • Hortonworks Administration Certification
- • MicroStrategy Certified Engineer
- • Talend Certified Administrator