Aman Kedia

Engineering Manager

Bengaluru, Karnataka, India9 yrs 6 mos experience
Most Likely To SwitchAI Enabled

Key Highlights

  • Led cross-functional teams at Amazon for tax compliance solutions.
  • Developed state-of-the-art NLP models achieving high accuracy.
  • Published multiple research papers on data mining and healthcare.
Stackforce AI infers this person is a Machine Learning and Software Development expert in Fintech and SaaS industries.

Contact

Skills

Core Skills

Artificial IntelligenceSoftware DevelopmentProject ManagementCloud ComputingMachine LearningNatural Language ProcessingData Science

Other Skills

AI-driven classificationtax compliancereal-time servicescross-functional team leadershipInvoicing platform developmentAWS OpenSearch migrationtask prioritizationsprint planningAWS OpenSearchdata migrationcost reductionrecommendation strategiesdata pre-processingalgorithm developmentdata analysis

About

As George Box once said 'all models are wrong, but some are useful', I am among those who are trying to find these useful models. I am a data enthusiast and lifelong learner. I enjoy solving the problems, whether they are small or big. I am an avid believer in Artificial Intelligence and algorithms supporting it. I have an interest in data science, health care, and languages and I am exploring how these fields are interdependent. Currently, I am working on Named Entity Recognition problems. The past year has been a great learning experience. I worked on some State of the Art problems in Natural Language Processing and established core components such as Kafka for data pipeline in my organization. I interacted with a lot of technology enthusiasts as well, delivering a few sessions on Data Science. I have worked at Oracle and SAP trying to solve problems leveraging the advancements in Artificial Intelligence. Prior to that, I worked at ISRO as an intern on detecting Disease patterns using Machine Learning on Time Series data and consequently clustering similarly behaving regions in Uttarakhand, India. I have 4 published research papers to my credit primarily on the amalgamation of Data Mining and Healthcare. 2 of my other works, also on Healthcare and Machine Learning have been submitted for publication. I use Python for most of my work since it is my go-to language for Data Science. I am fluent in core concepts around Java, Micro-services, Distributed Systems, and some other technologies. I completed my Masters from Birla Institute of Technology and Science in Computer Software Engineering. Please feel free to reach out on LinkedIn. Also, you can connect with me on Gmail: amankedia.sap@gmail.com Or my number: +91 9611853031

Experience

9 yrs 6 mos
Total Experience
3 yrs 2 mos
Average Tenure
6 yrs 3 mos
Current Experience

Amazon

3 roles

Software Development Manager

Promoted

Feb 2024Present · 2 yrs 4 mos

  • Team: Finance Technology - Building solutions for ensuring Amazon's tax compliance globally
  • Leading a cross-functional team of Applied Scientists and Software Development Engineers, driving roadmap definition, execution excellence, operational health, and long-term architecture for global tax compliance systems.
  • Owning Direct Tax Withholding for all Amazon eCommerce transactions across Retail and AWS—operating high-availability, real-time services processing 10K+ TPS and 25B+ transactions annually representing $175B+ in GMS. Responsible for end-to-end tax calculation, withholding, seller-level attribution, and orchestration at Amazon scale.
  • Spearhead the next-generation tax categorization platform for vendor invoices by integrating AI-driven classification using a dual-LLM “judge–jury” architecture with self-hosted large language models. Improving classification accuracy, vendor compliance, and downstream financial processing efficiency globally.
  • Drive engineering and scientific excellence through sprint planning, prioritization, design reviews, on-call readiness, and setting quality/latency/availability SLAs for mission-critical tax systems that are foundational to Amazon’s global financial operations.
  • Partner with Tax, Finance, Legal, and multiple engineering orgs to align regulatory requirements with scalable technical implementations, ensuring Amazon remains compliant across diverse international jurisdictions.
AI-driven classificationtax compliancereal-time servicescross-functional team leadershipArtificial IntelligenceSoftware Development

Software Development Engineer 3

Jul 2022Feb 2024 · 1 yr 7 mos

  • Team: Creator Technology - Building payments solutions for Creators in 27 locales generating invoices worth >$1BN yearly
  • Managing a team of 5 members, defined the team’s roadmap for 2024 and led activities around task prioritisation, sprint planning and system health improvements.
  • Designed and led the development of an Invoicing platform which could trigger pay-outs to Creators instantly post validations, earnings and tax computation followed by generating invoices and payment history reports for Creators. This helped move away from 1 monthly pay-out to multiple flexi pay-outs and enabled Rewards programs such as Flat Fee and Gift Card incentives.
  • Defined and led a phase-wise strategy for the migration of Creator Payment History from legacy Elasticsearch to AWS OpenSearch with 0 downtime backfilling 350MN+ entries. This offloaded processes such as data backup, cluster resizing to AWS, improved maintainability and saved $220K infrastructure costs yearly. During this process, I identified major areas of data issue in Payment history and guided others to fix them correcting reporting by $1.2 MN for 9K creators.
Invoicing platform developmentAWS OpenSearch migrationtask prioritizationsprint planningSoftware DevelopmentProject Management

SDE II / MLE

Mar 2020Jul 2022 · 2 yrs 4 mos

  • Team: Sponsored Products Advertising - Building Machine Learning based recommendation strategies and widgets.
  • Created a widget showcasing Compatible wireless accessories on smartphone pages increasing revenue by 220bps. Built pipelines for data pre-processing, training and inferencing using WAR-BERT model. Devised an algorithm for smartphone title deduplication, reducing data for inferencing by 99% and saving $260K costs. Ads shown on the widget had 99.6% compatibility.
  • Worked with 15+ teams and led the design of Advertisements on Amazon Bazaar, a store on Amazon where customers could find products at wholesale prices similar to Meesho/Shopee.
  • Proposed and developed the Add to Cart button for sponsored widgets on Amazon Fresh to enable One-click addition of products to customer’s cart. This led to 34 and 49 bps increase in revenue and sales respectively.
  • Designed a widget on Amazon Fresh which showcases 200+ cross-category products on Detail Pages to help customers build baskets. The widget had >99% coverage and was served with a p99 latency of 100ms and 0% failure rate.
Machine Learningrecommendation strategiesdata pre-processingalgorithm developmentSoftware Development

Packt

Author, Hands-On Python Natural Language Processing

May 2019Mar 2020 · 10 mos

  • Author, Hands-On Python Natural Language Processing
  • Amazon Link: https://www.amazon.com/dp/B08BG5581Y

Oracle

Senior Member Of Technical Staff

Jul 2018Mar 2020 · 1 yr 8 mos · Greater Bengaluru Area

  • Team: Oracle Digital Assistant (ODA) part of Oracle Cloud Infrastructure (OCI) - Developing technology for PaaS and SaaS Chatbots
  • Worked on developing Natural Language Processing (NLP) based Named Entity Recognition(NER) models for extracting Date, Time, Set, Duration and Email entities from chatbot utterances.
  • Studied Google BERT architecture and developed models for NER Detection and Classification using pre-trained embeddings from BERT. Achieved 99.1% accuracy on CoNLL 2003 dataset.
  • Explored possibilities of Universal Sentence Encoder for fetching sentence level embeddings.
  • Developed REST endpoints and backend logic for building an Entity Annotation pipeline.
  • Proposed, designed and established distributed Kafka architecture with scale-out capabilities for ODA message pipeline. It handles as high as 75 Mbps throughput on the producer end and 200+ Mbps throughput on the consumer end.
  • Developed Kubernetes Cluster monitoring capabilities for Kafka, Zookeeper, and Elasticsearch.
Natural Language ProcessingNamed Entity RecognitionKafka architectureSoftware Development

Sap

2 roles

Scholar | Developer | Machine Learning | Full Stack

Promoted

Apr 2017May 2018 · 1 yr 1 mo

  • Team: SAP Leonardo Machine Learning, Innovation Center Network
  • Researched and developed new Machine Learning and Deep Learning based models for the ranking resumes with respect to job details using Supervised and Unsupervised algorithms.
  • Improved the accuracy of the baseline by more than 10% by reading on the latest research on text mining, incorporating Embeddings from Word2Vec, enhancing the pre-processing techniques being employed and adding quality and relevant data.
  • Developed RESTful services from scratch for the team which enables to compare and evaluate the performance of multiple Machine Learning models on data of users’ choice.
  • Established pipelines for Machine Learning model training covering data collection, pre-processing and preparation, training, validation, and testing/prediction.
Machine LearningDeep LearningRESTful servicesSoftware Development

Scholar | Developer | ABAP | SAP ERP

Sep 2016Mar 2017 · 6 mos

  • Team : Fashion Management Solutions
  • Skills : ABAP, ABAP OO, ABAP Units
  • Modules worked on : Sales & Distribution (SD) & Materials Management (MM)
ABAPSAP ERPSoftware Development

Isro - indian space research organisation

Intern

Dec 2015Mar 2016 · 3 mos · Dehradun, India

  • I worked under Dr. Sameer Saran and Prof. Shiva Reddy Koti on the project entitled "Data Mining for Public Health Data for Dehradun".

Education

Birla Institute of Technology and Science, Pilani

M.Tech — Computer Software Engineering

Jan 2016Jan 2018

Institute Of Engineering and Management

Bachelor's Degree — Computer Science

Jan 2012Jan 2016

Stackforce found 100+ more professionals with Artificial Intelligence & Software Development

Explore similar profiles based on matching skills and experience