Aditya Pandey profile photo 1 on Tal

Aditya Pandey

Senior Software Engineer at Tekion | Ex-Amazon

Bengaluru

7+ years of work experience

Tekion Corp
Jul 20251 yr 2 mosPresent

Bangalore Urban, Karnataka, India

Architected and led the Data Sharing Platform — four microservices (reactive gateway, identity, data plane, freshness) enabling 40+ external partners to pull governed Delta tables from Unity Catalog (Databricks); drives ~$120K/month revenue on ~$11K/month infrastructure (~10× revenue-to-cost) and replaced a prior system costing ~$95K/month, an 88% cost reduction. Built production GenAI systems on the Anthropic SDK and Model Context Protocol (MCP): a multi-model analytics agent with a 2-tier hierarchical context-memory architecture (47% accuracy improvement) and an agent-based automation platform that cut service-onboarding turnaround from 2+ weeks to under 8 hours. Designed the manifest pipeline at the core of the data plane — background Delta-snapshot pre-compute (Delta Kernel, no Spark on the request path) for sub-second P50 reads, with an idempotent, distributed-lock-protected queue and fast-path cache; reused the same snapshot primitive to ship a customer-facing Data Freshness service with no duplicate reader. Delivered Kafka/SQS-class queueing — at-least-once delivery, consumer-group reclaim, and dead-letter handling — on the Redis the platform already ran, deliberately avoiding a new messaging fleet to keep infrastructure cost and operational overhead low. Built the authentication and multi-tenancy model: zero-knowledge SRP-6a auth (AWS Cognito) so partner credentials are never transmitted, with a fully offline-identical local/CI mode; hierarchical rate limits, per-tenant data isolation, and tag-based region overrides that kept EU data resident and unlocked international customers with no code changes. Reviewed 100% of PRs across all four services and mentored engineers and interns via design-first scoping; standardized Trino/EMR bootstrap (1 week → 1 day), multi-architecture Docker, and Smithy-first API contracts across the data team.
Amazon
May 20214 yrs 3 mosJul 2025

Built a Java-based orchestration manager on AWS Lambda coordinating 10K+ pipelines and 1B+ execution steps/month across Airflow and AWS Step Functions using an extensible, pluggable architecture. Designed large-scale ETL systems in Python processing 100+ PB/day with fault-tolerant workflows, supporting $7B+ in business operations. Architected a scalable Data Lake platform using Java (Spring Boot) and AWS, managing 1,200+ tables and 1M+ partitions with versioned historical data for analytics and regulatory compliance. Co-developed a Resource Manager microservice on AWS EMR, optimizing compute allocation for 35K+ executions/day across 20+ AWS accounts and eliminating capacity bottlenecks. Led a Change Impact Evaluation platform in Python with A/B testing, reducing deployment time by 93% (6 weeks → 2 days). Spearheaded CI/CD-driven global expansion of 30+ Java microservices across 40+ AWS regions using AWS CDK (Java & TypeScript), provisioning services, APIs, alarms, and dashboards. Built Data Discovery and Data Quality platforms (Java, Python) supporting 20K+ daily ELT executions and monitoring 35K+ runs/day with AWS Kinesis, RDBMS, and CloudWatch; designed SLA monitoring across 800+ tables preventing $4MM+ hourly impact.
Amazon Payment Services
Jul 20191 yr 10 mosApr 2021

Chennai

- Designed and implemented a AWS Lambda/EC2/S3/RDBMS/SQS/DynamoDB backed micro-service framework for Tier 1 services validation (with help of Selenium and TestNG), supporting 10+ applications and reducing manual testing time from over a month to ~10 days - Built multi-tenet plug and play framework to support 3+ browsers and 5+ operating systems for application testing - Owned the process and mechanism for enabling continuous integration and deployment (CI/CD) across 4+ service pipelines to prevent $10MM+ business impact - Invented a Cucumber-based framework integrated with TestNG and TestRail to reduce test case publishing time from 5 days to 30 minutes - Led a team of 2+ engineers to adopt the framework, delivering 450+ automated test cases and improving testing efficiency for 10+ teams - Developed a validation platform for applications and AWS services (SQS, Lambda) covering 100+ user stories using Java and design patterns, ensuring comprehensive backend and end-to-end functionality - Collaborated with 4+ teams to design the platform, enabling automation of over 100 user stories and 350+ test cases for both happy and negative scenario validation - Created 25+ reusable interfaces and modules and built a multi-tenant framework with Appium and TestNG for iOS and Android to validate application over 20+ devices, showcasing adaptability to new technologies
Amazon
Sep 20189 mosMay 2019

Greater Chennai Area

- Created a micro-service framework for end-to-end web validation, covering 400+ test cases and 150+ user stories - Integrated the framework with 15+ REST APIs to reproduce user behavior and improved validation success rate by 25+% with retry mechanisms - Simplified adoption of the framework by providing clear documentation and boosting usage by 60%. Reduced test case automation time from 2 days to 4 hours
GitHub
@pandeyad

Aditya’s AI Stack

anthropic apis
anthropic apis
antigravity
antigravity
claude code
claude code
claude
claude
codex
codex

Education

SRM IST Chennai

BTech - Bachelor of Technology, Computer Science

SRM IST Chennai

20154 yrs2019
army public school

High School Diploma

army public school

Join Aditya and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Aditya Pandey on Tal

Headline

Senior Software Engineer at Tekion | Ex-Amazon

About

At Amazon, I have specialised in designing and implementing robust, scalable orchestration services, directly contributing to the 10-fold increase in system efficiency. With a Bachelor of Technology in Computer Science from SRM University, my foundation in technical principles has been essential in crafting API models and leveraging AWS services for seamless ETL pipeline management. The passion for work and continuous skill development is manifest in our delivery of automated testing frameworks and CI/CD mechanisms, which significantly improve business processes and reduce manual efforts. This dedication to innovation and excellence is what propels me toward seeking advanced roles where I can further drive impactful solutions in software development.

Location

Bengaluru

AI stack

anthropic apis

antigravity

claude code

claude

codex

Work experience

Senior Software Engineer at Tekion Corp, Jul 2025 - Present · 1 yr 2 mos, Bangalore Urban, Karnataka, India, Architected and led the Data Sharing Platform — four microservices (reactive gateway, identity, data plane, freshness) enabling 40+ external partners to pull governed Delta tables from Unity Catalog (Databricks); drives ~$120K/month revenue on ~$11K/month infrastructure (~10× revenue-to-cost) and replaced a prior system costing ~$95K/month, an 88% cost reduction. Built production GenAI systems on the Anthropic SDK and Model Context Protocol (MCP): a multi-model analytics agent with a 2-tier hierarchical context-memory architecture (47% accuracy improvement) and an agent-based automation platform that cut service-onboarding turnaround from 2+ weeks to under 8 hours. Designed the manifest pipeline at the core of the data plane — background Delta-snapshot pre-compute (Delta Kernel, no Spark on the request path) for sub-second P50 reads, with an idempotent, distributed-lock-protected queue and fast-path cache; reused the same snapshot primitive to ship a customer-facing Data Freshness service with no duplicate reader. Delivered Kafka/SQS-class queueing — at-least-once delivery, consumer-group reclaim, and dead-letter handling — on the Redis the platform already ran, deliberately avoiding a new messaging fleet to keep infrastructure cost and operational overhead low. Built the authentication and multi-tenancy model: zero-knowledge SRP-6a auth (AWS Cognito) so partner credentials are never transmitted, with a fully offline-identical local/CI mode; hierarchical rate limits, per-tenant data isolation, and tag-based region overrides that kept EU data resident and unlocked international customers with no code changes. Reviewed 100% of PRs across all four services and mentored engineers and interns via design-first scoping; standardized Trino/EMR bootstrap (1 week → 1 day), multi-architecture Docker, and Smithy-first API contracts across the data team.

Senior Software Engineer at Amazon, May 2021 - Jul 2025 · 4 yrs 3 mos, Built a Java-based orchestration manager on AWS Lambda coordinating 10K+ pipelines and 1B+ execution steps/month across Airflow and AWS Step Functions using an extensible, pluggable architecture. Designed large-scale ETL systems in Python processing 100+ PB/day with fault-tolerant workflows, supporting $7B+ in business operations. Architected a scalable Data Lake platform using Java (Spring Boot) and AWS, managing 1,200+ tables and 1M+ partitions with versioned historical data for analytics and regulatory compliance. Co-developed a Resource Manager microservice on AWS EMR, optimizing compute allocation for 35K+ executions/day across 20+ AWS accounts and eliminating capacity bottlenecks. Led a Change Impact Evaluation platform in Python with A/B testing, reducing deployment time by 93% (6 weeks → 2 days). Spearheaded CI/CD-driven global expansion of 30+ Java microservices across 40+ AWS regions using AWS CDK (Java & TypeScript), provisioning services, APIs, alarms, and dashboards. Built Data Discovery and Data Quality platforms (Java, Python) supporting 20K+ daily ELT executions and monitoring 35K+ runs/day with AWS Kinesis, RDBMS, and CloudWatch; designed SLA monitoring across 800+ tables preventing $4MM+ hourly impact.

Software Development Engineer at Amazon Payment Services, Jul 2019 - Apr 2021 · 1 yr 10 mos, Chennai, - Designed and implemented a AWS Lambda/EC2/S3/RDBMS/SQS/DynamoDB backed micro-service framework for Tier 1 services validation (with help of Selenium and TestNG), supporting 10+ applications and reducing manual testing time from over a month to ~10 days - Built multi-tenet plug and play framework to support 3+ browsers and 5+ operating systems for application testing - Owned the process and mechanism for enabling continuous integration and deployment (CI/CD) across 4+ service pipelines to prevent $10MM+ business impact - Invented a Cucumber-based framework integrated with TestNG and TestRail to reduce test case publishing time from 5 days to 30 minutes - Led a team of 2+ engineers to adopt the framework, delivering 450+ automated test cases and improving testing efficiency for 10+ teams - Developed a validation platform for applications and AWS services (SQS, Lambda) covering 100+ user stories using Java and design patterns, ensuring comprehensive backend and end-to-end functionality - Collaborated with 4+ teams to design the platform, enabling automation of over 100 user stories and 350+ test cases for both happy and negative scenario validation - Created 25+ reusable interfaces and modules and built a multi-tenant framework with Appium and TestNG for iOS and Android to validate application over 20+ devices, showcasing adaptability to new technologies

Software Engineer Intern at Amazon, Sep 2018 - May 2019 · 9 mos, Greater Chennai Area, - Created a micro-service framework for end-to-end web validation, covering 400+ test cases and 150+ user stories - Integrated the framework with 15+ REST APIs to reproduce user behavior and improved validation success rate by 25+% with retry mechanisms - Simplified adoption of the framework by providing clear documentation and boosting usage by 60%. Reduced test case automation time from 2 days to 4 hours

Education

BTech - Bachelor of Technology, Computer Science, SRM IST Chennai

High School Diploma, army public school

Public links

github: https://github.com/pandeyad