RESUME

Kanja Saha

Staff Data & Analytics Engineer — AI-Ready Data Foundations & Agentic Analytics

Staff Data & Analytics Engineer with a proven track record of building data and reporting foundations from the ground up. I lead analytics projects with cross-functional teams to align on success metrics and set technical standards across the full data stack, from ingestion and semantic modeling to conversational and agentic analytics, empowering business users to make confident, data-driven decisions. I partner with the Machine Learning & Data Science team to build predictive models, RAG pipelines, prompt reasoning, and Guardrails, and to embed conversational analytics directly into existing dashboards. With 10+ years of experience across startups, enterprise, and consulting in B2B SaaS, I bridge the decision gap with data for accountable business impact.

Technical Skills & Certifications

Languages: SQL (CTEs, window functions, performance tuning), Spark SQL, Python (Boto3, pandas, scikit-learn)

Data Engineering: dbt (modeling, macros, unit tests, CI/CD), Airflow, MetricFlow, Git, IaC (Terraform, CloudFormation)

Data Modeling: Ontology, medallion architecture, dimensional modeling (Kimball), semantic layer design

AWS Platforms: Redshift, Athena, Glue, Kinesis, Lambda, S3, CloudWatch, SageMaker, Bedrock Studio, AgentCore

GCP Platforms: BigQuery, Cloud Storage, Pub/Sub, Dataflow, Dataproc, Cloud Monitoring

GenAI & ML: Databricks, Snowflake, Claude Code, Codex, RAG pipelines, Guardrails, Foundation Models

Visualization: Amazon Q, QuickSight, Tableau, Looker, Metabase, Hex

Data Science: Segmentation, churn prediction, propensity modeling, forecasting, experimentation, attribution

Certification: AWS Certified Machine Learning Specialty (2021)

Key Impact

• Architected data contracts and governance framework, aligning success metric definitions across business teams.
• Built a semantic layer, RAG pipelines, and contextual data foundations enabling AI-augmented analytics.
• Built a dbt agent for AWS Bedrock's 1T+ telemetry platform, cutting incident resolution time from hours to minutes.
• Built a GTM Agent with churn prediction models, automating at-risk detection and tailored outreach, reducing churn.
• Reduced reporting overhead by 80%+ through self-serve conversational analytics across 30+ stakeholders.
• Optimized AI workload utilization across 37 regions and 6 geographies, surfacing capacity and cost opportunities.
• Built 1Password's first data platform, incorporating finance, marketing, and sales data through the customer lifecycle.
• Designed multi-touch attribution models for a $1.13B ad business, driving 10% upsell and reducing churn by 5%.

Professional Experience

Author, Open Source and AI Agent Developer

May 2026 — Present

Authored Bridging the Gap with Data, built three production AI agents and an open-source dbt project drawing on 10+ years of hands-on analytics engineering experience.

Amazon Web Services — Staff Analytics Data Engineer, Amazon Bedrock Core Services

June 2024 — April 2026, Santa Clara, CA

Led analytics engineering for AWS Bedrock during a high-growth scale-up, defining the data architecture, modeling standards, and governance frameworks that enabled trusted decision-making across Product, GTM, Finance, and Engineering.

• Engineered scalable ingestion pipelines using AWS Glue, Kinesis, and S3 to process 1T+ raw telemetry records, establishing the foundational dbt Medallion Architecture, testing standards, and CI/CD practices.
• Architected data contract standards, including schema definitions, freshness SLAs, and quality gates, improving the reliability of downstream models as the platform scaled.
• Built the semantic layer, RAG pipelines, and contextual data foundations enabling AI-augmented analytics across Product, GTM, Finance, and Engineering.
• Built the semantic metrics layer that standardized KPI definitions across five functions, reducing reporting overhead by 80%+ for 30+ stakeholders.
• Delivered executive reporting that translated product telemetry into S-team and board-level narratives.
• Designed usage and capacity analytics across 37 regions and 6 geographies, surfacing GPU constraints that informed reallocation decisions and protected spend.
• Launched idle-instance monitoring for GPU compute, identifying underutilized capacity and improving infrastructure efficiency.
• Built a dbt Pipeline Health Agent using Amazon Bedrock and AgentCore to autonomously root-cause dbt test failures and pipeline bottlenecks, cutting incident resolution time from hours to minutes.

Amazon Web Services — Lead Analytics Engineer, Amazon SageMaker Canvas

March 2021 — May 2024, Santa Clara, CA

Joined SageMaker Canvas before meaningful usage data existed and built the analytics foundation from the ground up, defining how product, GTM, and Finance measured success.

• Engineered data pipelines to ingest cross-region product telemetry and financial data, building the Kimball-style data marts that became the foundation for all subsequent Canvas analytics.
• Architected the first end-to-end revenue data mart spanning ARR, pipeline, and goal tracking across 10+ SageMaker services; reconciled with Finance, adopted for WBR reporting, and reduced manual effort by 80%.
• Designed the experimentation and measurement framework for a Low Code No Code ML product, using usage data across 16 AWS regions to identify adoption friction, reduce support tickets by 10%, and improve time-to-first-model.
• Built ML segmentation models (Spark/scikit-learn) across 1,300+ enterprise accounts and architected a human-in-the-loop GTM Agent that flagged usage drops and drafted retention emails, reducing churn rate by 25%.
• Automated analytics for 5+ GTM campaigns using Salesforce pipeline and campaign data, improving enterprise win rate by 10% and shortening conversion time from 6 months to 3 months.
• Drove roadmap prioritization through usage analysis and performance benchmarking, influencing 10 features that improved customer milestone completion by 10%.

1Password — Senior Analytics Engineer

September 2020 — February 2021, Toronto, Canada

Built 1Password's first centralized analytics foundation, connecting commercial and product data into a scalable reporting and decision-support layer.

• Engineered ELT ingestion pipelines to extract Salesforce, Marketo, and billing records, transforming raw source data into Silver, Gold, and semantic layers to power the first centralized Looker stack.
• Defined revenue and retention metrics for freemium-to-paid conversion funnels, enabling Marketing and Finance to optimize spend and forecast subscription revenue.
• Eliminated hours of recurring manual reporting by creating self-serve dashboards and standardized business logic for commercial stakeholders.

Google — Full Stack Analytics Consultant

January 2015 — August 2020, Mountain View, CA

Built modular analytical data foundations and high-impact measurement solutions that supported advertising, experimentation, and lifecycle growth decisions at YouTube.

• Engineered modular SQL pipelines within Google's internal framework, utilizing DAG-based dependency tracking and Git version control equivalent to modern dbt workflows.
• Designed multi-touch attribution models for a $1.13B advertising business, surfacing cross-sell opportunities that drove 10% upsell from Search to YouTube ads and reduced advertiser churn by 5%.
• Built ML-driven segmentation models across advertiser cohorts using Spark and scikit-learn, improving targeted outreach and increasing email response rate by 20%.
• Instrumented A/B tests across YouTube UI flows and SMB email campaigns, increasing advertiser activation by 12% and reducing unsubscribe rates by 18%.

Education

M.B.A., Marketing and Information Systems

Indian Institute of Management, Bangalore, India

B.S., Computer Science

National Institute of Technology, Warangal, India