02 License savings
$75Kannual savings
Snowflake submissions replaced paid provider licenses, saving about $75K annually.
BUILD / SCALE / TRUST
I make production
data systems faster,
simpler, and easier
to trust.
From messy data to reliable products, I build production data platforms with measurable business impact.
RAW DATABUILD →PROCESS →DELIVERREAL OUTCOMES
03 Performance
30%faster processing
Query and pipeline tuning made S&P Capital IQ processing 30% faster.
04 Automation
15-minutepayment automation
Python, Lambda, and SQS now import payments every 15 minutes.
Experience
TURNING DATA
INTO PROGRESS
A track record of building and operating data systems at scale, in business-critical environments.
Aug 2021 — Present
Singapore & USA
Data Engineer
Partners Group
Own systems and applications for the portfolio monitoring product group, lead technical delivery, and manage budgets, vendor relationships, project plans, and stakeholders.
Production work,
measured in outcomes.
I build data products and cloud systems from the data model through production operations, with measurable outcomes for analysts and business teams.
Replaced paid data submissions with a Snowflake platform.
Built an API intake and dbt medallion pipeline with Airflow (MWAA) data quality checks. The platform saves about $75,000 in annual licensing costs; I also steward its Collibra catalog.
Snowflake · dbt · Airflow · CollibraBrought 8 TB of music streaming data into investment analysis.
Led two engineers in the Luminate data onboarding, shaping its Snowflake architecture and model alongside Cortex Agents and an MCP layer. Analyst effort fell by more than 50%.
Snowflake · Data modeling · Cortex Agents · MCPModeled uneven ownership for a €300M fund commitment.
Partnered with business teams on excuse rights data models and allocation logic for the Q4 2026 onboarding and future deals.
Investment data · Allocation logic · Stakeholder partnershipOwned systems and applications for portfolio monitoring.
Managed budgets, vendor contracts, project planning, and stakeholders; mentored engineers and reviewed designs and code. Led a five-vendor RFP to select a digital boardroom builder for the founders and executive team.
Product ownership · Technical leadership · Vendor selectionCut a comparables pipeline's processing time by 30%.
Improved the S&P Capital IQ benchmarking pipeline with denormalization, materialized views, query plan tuning, and incremental loads.
SQL · Query planning · Incremental loadsMigrated a 3 TB SQL Server database to AWS Aurora.
Moved the on-premises workload and reduced annual licensing and hardware costs by approximately $25,000.
AWS Aurora · SQL Server · Cloud migrationConnected portfolio data to natural-language analysis.
Built a FastMCP server over portfolio-company databases and connected it to LLMs through EPAM AI DIAL, letting business users query and compare portfolio data in natural language.
FastMCP · EPAM AI DIAL · MCPCut feature delivery time by over 50%.
Designed an AI development harness using AWS AI-DLC and Cursor CLI, with custom Java, React, and data-quality skills. Fifteen engineers across four departments adopted it.
AWS AI-DLC · Cursor CLI · Java · React · Data qualityReplaced Excel-based NAV oversight with a production app.
Built a net asset value oversight application in one month with Claude Code, a Java backend, React frontend, AWS EC2, and RDS. It saves two FTEs per year.
Claude Code · Java · React · AWS EC2 · RDSRemoved manual entry from recurring payment imports.
Built a queue-based import service with Python, AWS Lambda, and SQS to process payments every 15 minutes.
Python · AWS Lambda · SQSEarlier
Graduate Research Assistant
Colorado State University · Jan 2021 — May 2021
Researched representation of Indigenous peoples in academia with the College of Business, contributing insights to the department's academic framework.
TrustOps
Scam triage pipeline
A current build exploring real-time classification and evaluation, with human review for uncertain or high-risk cases. View the code ↗
01 / ModelFine-tuned Qwen2.5-0.5B with LoRA to classify scam type and risk level, measured against a rules baseline on macro F1 and high-risk recall.
02 / StreamStreams cases through Kafka to a typed FastAPI scoring service, routing invalid or high risk output to human review.
03 / QualityUses reproducible SHA-256 data splits, schema validation, and CI that tests and evaluates every push.
Python · Apache Kafka · Hugging Face Transformers · TRL · PEFT · FastAPI · Pydantic · pytest · Docker · GitHub Actions
The system behind
the outcomes.
Tools matter when they support a sound engineering decision. These are the areas I use to move from diagnosis to a stable production result.
INGESTMODELOPERATEIMPROVE
- Data engineering
- Snowflake, dbt, Python, SQL, PySpark, Kafka, Airflow, data modeling, ETL, medallion architecture
- Cloud platform
- AWS S3, Lambda, SQS, EC2, RDS, Aurora, Redshift, Athena, MWAA, CloudWatch
- AI systems
- MCP, FastMCP, Cortex Agents, semantic views, EPAM AI DIAL, AWS AI-DLC, Cursor CLI, Claude Code, FastAPI, LoRA, model evaluation
- Delivery
- Terraform, Docker, GitHub Actions, Collibra, Java, React, Streamlit, Tableau
Credentials,
open for verification.
Each certificate links to the issuer record. No screenshots, no unverifiable badge wall.
Education
Technical depth with a business systems perspective.
Master of Computer Information Systems
Colorado State University · Graduate certificates in Business Intelligence and Application Development
Bachelor of Engineering in Information Technology
University of Mumbai