Hire Me →
Available for Hire

Data
Architect.

Architecture that scales. Costs that don't.

_
4× Databricks Certified 2× Azure Certified 1× Fabric Certified
View Projects Hire Me →
25+
Projects
6yr+
Experience
8+
Clients
25TB+
Data Processed
dach@cloud-prod:~$
847,293 rows/sec
12 pipelines
99.97% uptime
2.41 TB today
// LIVE PIPELINE STATUS ALL SYSTEMS NOMINAL
ADF
Databricks
Delta Lake
Fabric
Power BI
AzureDatabricksMicrosoft FabricOneLakeDelta LakeLakehouse ArchitectureHybrid Data ArchitecturePySparkSpark SQLPythonSQLScalaAzure Data FactoryFabric PipelinesDataflow Gen2dbtApache AirflowApache KafkaAzure Event HubsADLS Gen2Unity CatalogPower BIDAXSemantic ModelsTerraformDockerAzure DevOpsGitHub ActionsCI/CDManaged IdentityKey VaultPurviewData QualityObservabilityGrafanaCost OptimizationData GovernanceLineage AzureDatabricksMicrosoft FabricOneLakeDelta LakeLakehouse ArchitectureHybrid Data ArchitecturePySparkSpark SQLPythonSQLScalaAzure Data FactoryFabric PipelinesDataflow Gen2dbtApache AirflowApache KafkaAzure Event HubsADLS Gen2Unity CatalogPower BIDAXSemantic ModelsTerraformDockerAzure DevOpsGitHub ActionsCI/CDManaged IdentityKey VaultPurviewData QualityObservabilityGrafanaCost OptimizationData GovernanceLineage

// By the Numbers

Impact That Speaks

$10→ $0.02
Pipeline Cost Per Run

500x cost reduction through compute optimization, orchestration tuning, and intelligent workload scheduling.

70%→ 99.99%
Pipeline Reliability

Improved success rate with retry frameworks, proactive monitoring, and automated alerts.

25TB+
Data Processed

High-throughput lakehouse pipelines across historical datasets and 100GB/day financial ingestion.

40 min→ 4 min
Dashboard Refresh Time

10x faster reporting refresh by improving transformations, aggregation logic, and semantic-layer readiness.

45+→ 1
Orchestration Simplified

Reduced a multi-step notebook chain to one controlled trigger with run-level logging and traceability.

6yr+
Tech & Data Experience

Finance · Insurance · Telecom · Healthcare · AI/ML.

Trusted across Financial Services Insurance Telecom Healthcare AI / ML Platforms Embedded Systems

// What I Do

Core Services

Design, build, modernize, and stabilize data platforms from source systems to decision-ready analytics.

Platform Architecture & Strategy

Define the right architecture, roadmap, standards, and delivery plan before teams commit time and budget.

Architecture Roadmap Standards

Pipeline Build & Automation

Build reliable data movement, transformation, scheduling, and workflow automation for production use.

ETL / ELT Automation Orchestration

Cloud & Legacy Modernization

Move outdated scripts, desktop reporting flows, and fragmented pipelines into maintainable modern data platforms.

Cloud Migration Lakehouse Refactor

Analytics & Reporting Layer

Turn operational data into clean models, trusted metrics, and dashboard-ready datasets for business users.

Data Modeling Metrics Layer BI Backend

Reliability, Cost & Governance

Improve data trust, performance, spend, monitoring, access controls, and operational visibility.

Data Quality Cost Monitoring

AI & ML Data Foundations

Prepare dependable data flows for machine learning, search, classification, RAG, and intelligent products.

ML Pipelines RAG Data Feature Prep

// Technical Stack

Skills & Expertise

// Data Engineering

Apache Spark / PySpark95%
Python / SQL / Scala95%
Delta Lake / Lakehouse Architecture95%
Azure Data Factory / Fabric Pipelines93%
Airflow / Kafka / dbt85%

// Platforms & Delivery

Databricks (4x Certified)97%
Microsoft Azure (2x Certified)93%
Microsoft Fabric (1x Certified) / OneLake90%
Terraform / Docker / Azure DevOps82%
Power BI / Metabase / Grafana85%

// Featured Work

Case Studies

Selected work across platform modernization, financial analytics, reliability engineering, AI-assisted workflows, and secure data operations.

Migration & Modernization

Legacy Margin Analytics to Modern Lakehouse

Modernized legacy ERP extracts and desktop database/forms-based margin reporting into a cloud lakehouse, preserving complex financial logic while making the pipeline easier to validate and operate.

  • Delivered 12+ Silver transformations, 12 analytical views, and 3 Gold fact tables.
  • Preserved cost hierarchy, cascade-style inheritance, customer overrides, special cost adjustments, and client pricing logic.
Outcome: Power BI-ready reporting with clearer lineage and stronger auditability.
Microsoft FabricMedallionPower BI
Reliability & Cost

Enterprise Pipeline Reliability Upgrade

Improved production lakehouse pipelines by tightening orchestration, retries, monitoring, and workload scheduling across high-volume financial data workflows.

  • Raised pipeline reliability from roughly 70% success rate to 99.99% uptime.
  • Reduced execution cost from $10 to about $0.02 per run through tuning and scheduling.
Outcome: lower operating cost, fewer failed runs, and stronger production confidence.
ADFDatabricksMonitoring
Platform Modernization

Azure Data Stack to Fabric Lakehouse

Led lakehouse modernization into Microsoft Fabric, introducing medallion architecture, OneLake, Dataflow Gen2, and cleaner reporting paths for analytics teams.

  • Reduced platform costs by around 30% after migration and architecture cleanup.
  • Improved dashboard load time from about 1 hour to under 3 minutes.
Outcome: faster analytics delivery on a cleaner modern data platform.
OneLakeDataflow Gen2Lakehouse
Data Quality & Auditability

Financial Logic Validation Framework

Standardized join handling and cost lineage across Silver, View, and Gold layers to reduce missed matches and make business-rule defects easier to trace.

  • Normalized 20+ notebook and view join paths across customer, item, group, and date keys.
  • Reworked weighted pricing and branch-level cost logic for accurate multi-customer reporting.
Outcome: fewer data defects, clearer debugging, and more trustworthy margin results.
SQLValidationAuditability
AI Workflow Automation

NLP Document Classification Pipeline

Optimized document classification workflows using NLP matching and LLM-assisted processing to reduce manual review effort and improve operational visibility.

  • Reduced manual policy classification from 51% to 0.3%.
  • Improved processing performance by 90%, cutting 5-7 minute runs to under 60 seconds.
Outcome: faster classification, less manual triage, and more scalable document workflows.
NLPLLMSpark
Observability & Secure Ingestion

Operational Visibility for Data Platforms

Built monitoring, alerting, and secure ingestion patterns for business-critical pipelines handling confidential financial data and large daily ingestion volumes.

  • Built pipeline health dashboards using operational metrics for leadership visibility.
  • Implemented secure ingestion patterns for confidential datasets using encryption controls.
Outcome: stronger governance, faster incident response, and safer production operations.
ObservabilitySecurityGovernance

// How I Work

From Problem to Production.

01
Discovery
Understand the data landscape, business context, and pain points
02
Architecture
Design the target state with ADRs, data models, and cost guardrails
03
Build
Iterative pipeline development with validation and testing at each sprint
04
Validate
Reconciliation checks, data quality rules, and stakeholder review
05
Handoff
Full documentation, runbooks, recorded walkthroughs, and team pairing
06
Support
Post-launch stabilization, monitoring, and ongoing optimization

// Products

Tools I've Built.

PDF Splitter

● Live

Split any PDF into smaller files instantly. Drag & drop interface, custom page-per-file selector, automatic ZIP packaging. Works completely offline — your files never leave your computer.

Desktop App Python MSIX Microsoft Store
✦ Drag & drop ✦ Custom page splits ✦ Save as ZIP ✦ 100% offline ✦ Dark theme UI ✦ Free & no ads
Open Tool →

Sight

● Live

A universal AI command bar for the modern enterprise desktop. One lens, every role, zero friction — from security guards to CEOs, everyone opens the same tool, types what they need, and gets things done.

Desktop App Tauri React + TypeScript Ollama TailwindCSS Framer Motion
✦ Universal command bar ✦ Role-based AI assist ✦ Local LLM runtime ✦ No app-hopping ✦ Enterprise-ready ✦ 100% offline

More Coming Soon

● In Progress

More useful tools are in progress, focused on everyday workflow problems, clean interfaces, fast outputs, and privacy-friendly experiences where possible.

Workflow Tools Utilities Automation Offline First

// Blog

Field Notes on Data Architecture.

// Engagements

Start With Scope, Not Guesswork

Strategic Platform Assessment
Discovery first
For unclear scope, reliability risks, or modernization planning
A senior review of your current data platform, reporting flow, quality gaps, and delivery risks before committing to a build.
  • Current-state architecture review
  • Pipeline reliability and failure points
  • Reporting logic and data quality risks
  • Cost and performance improvement areas
  • Prioritized roadmap and delivery estimate
Request Review
Recommended
Data Platform Delivery
Scoped build
For teams ready to modernize, rebuild, or launch production data workflows
Hands-on delivery for reliable pipelines, analytics-ready data models, validation checks, and operational handoff.
  • Target architecture and data model design
  • Pipeline build or modernization
  • Validation rules and reconciliation checks
  • Documentation, runbooks, and handoff
  • Stakeholder review and delivery checkpoints
  • Launch support and stabilization
Book Discovery
Fractional Data Architecture
Ongoing guidance
For teams that need senior architecture input without a full-time hire
Embedded architecture support for platform decisions, delivery reviews, standards, and long-term reliability guardrails.
  • Architecture decisions and technical reviews
  • Delivery roadmap and backlog shaping
  • Standards for modeling, testing, and observability
  • Cost and reliability guardrails
  • Team mentoring and stakeholder alignment
  • Monthly platform review
Discuss Fit

Pricing is scoped after a short discovery call because platform work depends on data volume, current stack, delivery urgency, compliance needs, and how much your internal team owns.

// About Me

Data Platform Architect.
Engineer. Builder.

Kush portrait
KK

Hi,
I’m Kush. I design and build data platforms for teams that cannot afford fragile pipelines, unclear reporting logic, or architectures that fail when the business grows.

My work sits where engineering discipline meets business pressure: modernizing legacy reporting flows, building lakehouse data platforms, standardizing transformation logic, and turning messy operational data into trusted analytics.

Across Finance, Insurance, Telecom, Healthcare, Embedded Systems, and AI/ML-driven platforms, I focus on systems that are accurate, explainable, cost-aware, and maintainable after handoff. Based in Pune, India, working with teams worldwide.

4x Databricks Certified 2x Azure Certified 1x Microsoft Fabric
Download Resume ↓ Work With Me →

// What People Say

Client Feedback

"

Kush rebuilt our entire reporting layer without a single day of downtime. He reverse-engineered business logic buried in 8-year-old stored procedures and turned it into clean, tested medallion models. Our finance team finally trusts the numbers.

MC
Michael Chen
VP of Analytics, Financial Services
"

We brought Kush in to fix pipeline failures that were costing us 3 hours every morning. He didn't just patch it — he redesigned the orchestration, added retry logic, and cut our per-run cost from $10 to under 2 cents. The guy thinks in systems, not scripts.

AS
Ankit Srivastava
Engineering Manager, Insurance Platform
"

What sets Kush apart is documentation. Every pipeline he delivers comes with lineage diagrams, cost breakdowns, and operational runbooks. When he handed off the Fabric migration, our team was self-sufficient from day one. Rare to find that in a contractor.

SR
Sarah Reynolds
Head of Data Engineering, Telecom

// Get In Touch

Start a Project

Let's build something great.

Share the business problem, current data setup, and what a successful outcome looks like. I will review the context and suggest the right next step before we talk scope.

Email

Kush@withdach.com

Response Time

Within 24 hours

Location

Pune, India — Remote Worldwide

Availability

Open for projects — Now

PLAY
// RETRO ARCADE
▶
Data Runner
Jump pipes, stomp goombas, collect coins
Space / Tap
Tetris
Classic block stacking, clear lines to score
Arrow Keys
Snake
Eat, grow, don't crash into yourself
Arrow Keys