Trust Before Throughput
If teams cannot trust the data, the rest is theatre. I care about timeliness, accuracy, completeness, correctness, and the operational signals that prove those things are true.
London
•
Booking.com
•
Lead Data Software Engineer
I engineer data systems that survive scale, audit, and change. Tushar Kesarwani
I work at the part of data engineering where architecture decisions matter: modern data platforms, cloud migration, data quality, governance, and distributed systems that still make sense six months after launch. Most of my work is about turning messy estates into calmer, more trustworthy systems.
Cloud-native and cloud-agnostic
AWS + GCP
Iceberg + Snowflake
Quality + Governance
How I Think
I do not treat data engineering as a pipeline factory. The interesting work is making large systems more trustworthy, more legible, and less dependent on heroics.
Architecture With Consequences
Cloud choices, storage models, governance boundaries, and platform abstractions all compound over time. I prefer designs that age well instead of ones that look clever for a sprint.
Leverage Over Noise
The best engineering often removes future work: stronger platforms, better defaults, cleaner migration paths, and automation that pays back across many teams.
Selected Work
A few examples that reflect the kind of problems I like: large-scale data architecture, difficult migrations, governance with real impact, and automation that creates engineering leverage.
Modern Platform
Booking.com's FinTech data estate
I have worked on cloud-agnostic data tooling, modern data architecture, and the migration of on-prem data workloads into the cloud. That includes a modern Iceberg data lake on AWS with quality and governance built in rather than bolted on later.
Built for petabyte-scale data assets with deeper
integration into Snowflake-backed workflows.
Data Reliability
Quality and governance as core product features
I helped build a data-quality microservice covering timeliness, accuracy, completeness, and correctness across on-prem and cloud assets. I also helped establish governance foundations that made the platform more accountable and easier to trust.
Quality and governance designed into the system, not added as
postmortem paperwork.
Real-Time Systems
Lloyds at transaction scale
Before Booking.com, I built real-time and batch Spark pipelines for Lloyds Bank, generating customer-facing notifications from high-volume transaction flows and helping scale structured streaming workloads on Kafka.
Supported 26M customers and around
100M daily transactions.
Engineering Leverage
Automation, hiring, and systems around the system
I won a 2024 hackathon by using GPT-based automation to convert Oozie workflows into Airflow DAGs, saving substantial developer time. I also built hiring playbooks, interviewed 300+ candidates, and helped build SOX lineage capability that avoided multi-million-dollar risk.
300+ interviews and an estimated
$4M in avoided fines through lineage work.
Toolbelt
The stack changes. The job does not: design dependable systems, make the data easier to reason about, and leave behind a platform that other engineers can actually use.
AWS
GCP
Scala
Python
Spark
Kafka
Snowflake
BigQuery
Iceberg
Delta Lake
Flink
DBT
Terraform
Kubernetes
Docker
CI/CD
Data Governance
Data Quality
Data Mesh
RAG
Earlier Chapters
Earlier in my career I worked across Sapient and Cognizant on ETL, analytics, migration, and big-data initiatives for advertising, banking, insurance, and e-commerce clients. That mix gave me a durable instinct for separating temporary complexity from structural complexity.