Manoj Sebastian

19+ years of experience in delivering data-driven platforms in the financial services domain, specializing in middle-office and client-facing data products, hands-on expertise across modern data stacks including AWS, Azure, Databricks, Spark, Python, Iceberg, Snowflake, Open AI, Open Router, RAG & Pinecone.

AI Programmer at Varsha-Info.com (2025-09 – Present)

Designed and developed an AI-native service marketplace that utilizes a Retrieval-Augmented Generation (RAG) architecture to transform how users discover local service providers. Engineered a sophisticated discovery engine that moves beyond keyword matching by using semantic search to interpret complex, natural language user intent.

• Implemented a vector database to index service provider metadata as embeddings, enabling high-precision semantic search that matches user queries based on context and intent rather than just keywords.
• Developed a Retrieval-Augmented Generation (RAG) pipeline where an LLM processes unstructured user requirements, retrieves relevant provider data from the vector store, and generates tailored service recommendations.

Data Warehouse [DW] Developer at State Street Corporation (2020-11 – 2025-09)

Working as a developer cum team lead in IMS (Investor Managerial Services). This system provides Middle Office data feeds to external clients from its record keeping system. This project follows agile methodology. While working on this system, it gives progress in technical skills, moreover, be familiar with middle office data streams like transactional, valuation & Reference.

• Developing data lineage from multiple data sources & aggregating it at later stage, to meet client data requirements.
• Working with Business Analysts to understand the data requirements.
• Data Modeling, development of data marts & client extracts.
• Performance tuning, Pre-Prod & Prod deployment.
• Experience in Agile methodology, working as SCRUM master.

Production Support Analyst at MUFG Investor Services (2018-08 – 2020-11)

Providing production support on various financial applications including Flexcube (Banking), Mantra (Investor Services), Geneva (Fund Accounting) etc. This provided good understanding about MUFG Application Paradigm, its interconnection.

• Addressing Business user requests related to data issues & application availability. Supporting data integration between different business applications.
• Performing End of Day (EOD) activity on different applications.
• Addressing performance issues in databases.
• Coordinating Major Incident calls to resolve critical issues.

Application Analyst at State Street Corporation (2011-07 – 2018-08)

Project - IMS SDO (Standard data offering) - State Street middle office system (IMS) consists of data warehouses which stores investment accounting details like daily valuation, transaction, cash position etc. This information belongs to external clients. There is an ETL application which will ensure that transaction, corporate actions, valuation data are loaded into data warehouses & deliver client specific extracts with target Client SLAs (Service Level Agreement).

Key role & responsibility to address exceptions including: 1) database performance –queries do not have proper index / tuned 2) System issues – network /hardware/storage issue 3) Data issue/volume – Incoming records have erroneous records, huge volume of records, backdated records may take more time to complete 4) Auxiliary system issues – Scheduling & password matrix system goes down 5) Issues related to recent deployments/release/patches etc.

ETL Developer on GPA data Integration at State Street Corporation (2007-07 – 2011-07)

GPA (Global Performance analytics) is the system which provides performance, attribution, and risk related services to clients. GPA will do specific functions like return calculation and dispersion analysis etc. on valuation records of funds. Design & implementation of ETL jobs (Data stage, PDI) to integrate external data into GPA data warehouse.

This involves modeling of a schema –IH (Integration Hub) with two set of tables 1- Dimensional & 2-Fact. While this module operates, static data has been loaded first to dimensional tables. Later, dynamic data has been loaded to fact tables while looking -up related attribute values in dimensional tables & enriched with key values.

In case of any data exceptions (data look up failures), kick out report will be generated for data validation. kick out report will be used by business to fix the data exceptions, correct them in the next run.

Production Support Analyst at State Street Corporation (2007-01 – 2018-08)

MCA in Master of Computer Application (2004)