Work
Director of Engineering at Kinesso, specializing in data architecture, cloud-scale analytics, and audience data platforms. I lead the teams and the architecture behind how brands understand and reach their audiences.
Career
Click the green dot for responsibilities
Director, Engineering
Lead the data engineering organization responsible for Kinesso's audience data platform, including the AI capabilities built on that data. Scope covers technical direction, delivery, team leadership, and cost across four core markets (US, UK, Germany, Australia) and 60+ regional markets.
Technical direction & architecture
- Guide the platform architecture for NeXus on Snowflake, including multi-region design and data-model evolution from NeXus 1.0 through 1.1 and 2.0.
- Stay engaged in the technical work closely enough to support sound architecture decisions and help resolve complex design questions.
- Led the data engineering organization through a multi-market platform transformation, migrating markets off legacy systems while continuing to deliver.
AI & agentic engineering
- Own the AI roadmap for the platform alongside core data engineering delivery.
- Designed and built Osiris, a multi-agent system that turns a plain-language brief into a validated audience through criteria extraction, entity-aware semantic search, boolean assembly, and composition and overlap checks, using LangGraph, Snowflake Cortex, vector search, and Container Services.
- Built NeXus Studio and NeXus Lab as Next.js applications connected to live Snowflake data: a structured audience workspace with semantic search, and a governed SQL notebook with metadata-aware results and query optimization support, developed with Cursor and multiple models.
- Established internal AI tooling for the data engineering team, including Claude Code skills integrated with Snowflake and Jira over MCP for operational workflows such as ticket creation, query review, account checks, and daily platform briefs.
- Encourage practical, responsible use of AI-assisted engineering tools across the team (Cursor, Claude, Cortex, and related tooling) as part of how we design and deliver.
Delivery & roadmap
- Own the data engineering roadmap and quarterly planning, translating priorities into epics and sequencing work across markets.
- Define and track platform outcomes across cost, performance, reliability, and impact on the audience product the business depends on, including AI surfaces where relevant.
- Accountable for delivery of the data backbone supporting B&AA, from intake through activation.
People & org
- Lead a distributed data engineering team of about 14 engineers across the US and Kuala Lumpur, including hiring, mentoring, and an onshore/offshore delivery model.
- Help engineers take clear ownership of domains such as ingestion, identity, warehouses, markets, and AI-on-data, with knowledge transfer and written decision records so the team can scale sustainably.
Stakeholders & partnership
- Work with product, client-facing teams, and peer engineering groups to align priorities and surface tradeoffs early.
- Operate across a multi-market, multi-partner environment; keep the data engineering and AI roadmap focused on the outcomes that matter most.
- Communicate platform status, risks, cost, and timelines clearly to leadership and partners.
- Act as a technical partner for peer teams when platform or Snowflake problems are hard to move past alone: help investigate, recommend a path, and leave them with something they can verify and own.
- Recent example: worked with Data Science to optimize a large People Universe build that was taking on the order of 15 hours; delivered a measured approach that finished in about 1.5 hours with identical results and lower cost, documented so their team could adopt it.
Cost, reliability & operations
- Own Snowflake and related platform cost and performance as data volume and market coverage grow.
- Maintain reliability at scale across markets, including migration and decommissioning of legacy systems as part of ongoing platform ownership.
Featured Projects
Architected and led Kinesso's next-generation audience data platform on Snowflake, replacing a legacy Elasticsearch system at scale. I drove the technical strategy, team execution, and stakeholder alignment across the full platform lifecycle.
NeXus spans 60+ markets, holds hundreds of terabytes, and resolves hundreds of millions of consumers. It powers B&AA, the product where client and sales teams build and activate audiences for pitches and live campaigns, with hundreds of client workspaces and tens of thousands of audiences built on top. That makes it revenue-linked infrastructure, not back-office plumbing. On top of the data layer I also built the Studio and Lab experiences that let people work the platform directly.
- NeXus 1.0 — Directed the core migration from Elasticsearch to Snowflake. The goal was to get off the legacy stack as fast as possible, so I adapted a data model that could reuse the data we had already built for Elasticsearch, which made for a quick migration. That move replaced a roughly $2M/yr always-on infrastructure cost base with elastic, pay-per-use Snowflake compute at about $1M/yr — roughly a 50% cost reduction — with query times up to 95% faster on certain workloads.
- NeXus 1.1 — Spearheaded the extension of the platform to support survey datasets, solving a fundamentally different join model at scale. Achieved a 25x reduction in data scanned on response-level data, with a matching ~25x speedup in query execution.
- NeXus 2.0 — Defined the architecture for a wide/column-based data model, eliminating upfront pre-materialization for complex fused datasets and enabling new categories of data providers.
- NeXus Studio — Built a structured audience workspace on Next.js over live NeXus 1 and NeXus 2 data: unified catalog with semantic search, nested boolean trees, universe-scoped sizing, save/export/report, and engine-transparent SQL. You drive the expression yourself; the count reacts as attributes land.
- NeXus Lab — Built a governed SQL notebook for analysts who write their own queries over NeXus 2 wide tables: logical views generated from metadata, schema-aware editing, results decoded from the taxonomy, and an optional Optimize step when a query needs to get faster.
Oversaw the design and delivery of a governed, API-driven ingestion service that handles end-to-end onboarding of first-party and third-party data into NeXus, from identity mapping and transformation through metadata registration and audience publication. Authentication runs on an OAuth layer built with AWS API Gateway and Cognito.
I also led a single-tenant deployment of SANDS that runs in fully isolated environments, built for first-party data with no PII in scope. It extends the platform to client setups the shared multi-tenant model was never meant to serve.
Led the data engineering team delivering the audience-building platform on Elasticsearch, an inherited architecture. Directed the pipeline and infrastructure optimization that kept it reliable at scale, cutting data-processing time by 40%. Recognized where Elasticsearch would break at higher volumes and made the case for the migration to Snowflake, which became the NeXus program.
AI & Agentic Engineering
Researched and built a multi-agent system that turns a plain-language brief into a finished, validated audience. It reads the brief, extracts the targeting criteria, runs entity-aware semantic search across the taxonomy, assembles a boolean audience tree, and returns geography, composition, and overlap validation on the result. Where Studio is a structured builder you drive yourself, Osiris hands the model a goal and a toolbox and lets it decide the steps.
I designed it to run inside Snowflake: LangGraph orchestration on Container Services, Cortex LLMs for extraction and reranking, vector embeddings for semantic search, Hybrid Tables for agent session state, and a Streamlit-in-Snowflake interface, with the same App Runtime / Next.js path the Studio experiences use.
Built the NeXus Studio and Lab front ends end to end in Cursor, working across multiple models in the same workflow rather than treating AI as autocomplete on the side. The surfaces are Next.js apps wired live to Snowflake: catalog browse and semantic search, an expression canvas with nested boolean trees, sizing telemetry and reports, plus a SQL Lab with metadata rails and decoded results. That is how a data platform team shipped a full product UI without waiting on a separate frontend track.
Led the creation of a suite of Claude Code skills the data team uses every day, wired to Snowflake and Jira over MCP. They automate the work that used to eat hours: standing up data-ops tickets, auditing query performance across the warehouse, running multi-account security and DBA checks, profiling tables, and producing a morning brief over Jira plus 24 hours of Snowflake health. The idea is to encode senior data-architecture judgment into tools the whole team can run.
Beyond the surfaces above, I stay fluent across Claude, GitHub Copilot, Lovable.dev when an idea needs to be seen rather than described, and Snowflake's Cortex Code (CoCo) for AI-assisted work right inside the warehouse. Staying current on the toolchain is part of how I keep technical judgment sharp.
Engineering Leadership
I lead a distributed engineering team spread across geographies, with engineers in the US and Kuala Lumpur, and I work closely with regional teams in Australia, the UK, Poland, and beyond as the platform serves their markets. Building one delivery model out of that spread was as much of the job as the architecture itself.
Day to day that means owning the roadmap and quarterly planning, breaking strategy into epics the team can actually execute, and investing in knowledge transfer so the platform never rests on a single person. For the bigger technical calls I run a written, meeting-free decision process where proposals get argued on paper before anyone commits, and I make the final call.
Earlier Work
Owned the delivery of an automated audience system on Redshift that served enterprise clients across retail, auto, and media. Config-driven pipelines and a lightweight DSL let teams stand up new client audiences without re-engineering each one, with access governance built in from the start. I also built the Scala and Spark encoding layer on AWS that made very large audiences fast to assemble, and integrated a wide range of third-party data sources behind it.
Before Cadreon, I came up through the Hadoop era of big data, working across MapReduce, Hive, Pig, Sqoop, HBase, and Cassandra, and stood up Hadoop on AWS while leading a delivery module. It is the foundation everything since has been built on.
Cloud & Infrastructure
I have built on AWS for over a decade, from standing up Hadoop clusters on EC2 in the early days to the event-driven, serverless data pipelines I design now. What I care about is how the pieces fit together: decoupled storage and compute, elasticity against cost, IAM boundaries, and governance across multiple accounts. I provision it as code with Terraform, so buckets, Lambdas, and configuration are reproducible and reviewed rather than clicked together by hand. Halving a roughly $2M/yr always-on footprint to about $1M/yr on pay-per-use compute was as much a cloud-economics call as an engineering one.
| Layer | Services |
|---|---|
| Compute & processing | EMR · EC2 · Lambda |
| Storage & data | S3 · Redshift · RDS · Glue |
| Streaming & orchestration | Kinesis · Step Functions · Managed Airflow (MWAA) |
| Analytics & search | Athena · OpenSearch · QuickSight |
| Integration & messaging | API Gateway · SNS · SQS · Transfer Family (SFTP) |
| Infrastructure as code | Terraform · CloudFormation |
| Monitoring & operations | CloudWatch · Storage Lens |
| Security & identity | IAM · Cognito |
Skills
Education
B.Tech in Computer Science & Engineering · Uttarakhand Technical University, India · 2009–2013
Writing on Medium
- Why Data Engineers Shouldn't Fear AI — They Should Embrace It Jun 2025
- Snowflake's Split Personality: Hybrid Tables vs Native — Who Wins What? Apr 2025
- Single vs. Multi-Tenancy in Snowflake: A Practical Breakdown for Data Architects Apr 2025
- How Data Engineering Powers Real-Time Decision-Making in Formula 1 Apr 2025
- Understanding Modern Data Architecture: From Databases to Lakehouses Apr 2025
- AI + Lovable.dev: The Fastest Way to Turn Product Ideas into Interactive Designs Apr 2025
- AWS OpenSearch: Promises vs. Reality — Why It Didn't Work for Us Mar 2025
On LinkedIn
I regularly share thoughts across technology, leadership, and broader ideas shaping the world. Pick a topic:
How human-AI collaboration drives breakthrough innovation — and why the best teams combine human judgment with AI speed and scale.
The technical stack behind production-ready AI agents — how Model Context Protocol and reusable skills enable real enterprise integrations.
Practical takes on AI-assisted engineering — from making tools like Claude Code accessible, to using AI as a collaborative partner for architectural thinking, optimization, and building reusable systems.
Deep dives into tools, tradeoffs, and real-world outcomes — from what changed in Apache Spark 4.0 to cutting Snowflake query costs by 75% using AI-assisted optimization.
Applying data science and AI to Formula 1 — ELO-based driver ratings, circuit-fit scoring, and race predictions updated throughout the season.
Takes on the big issues already in motion — AI's impact on employment, climate timelines, water scarcity, and biotechnology.