Ashwanth Kumar

CTO

Building signal-driven ad tech that leverages data from quick commerce platforms to improve campaign performance and drive smarter budget allocation.

Director & Fractional Software Architect

AK Labs is my personal company under which I provide consulting services to my clients. I wear multiple hats which includes:

• Building a pipeline of work
• Conducting sales calls with leads
• Managing finances for contractors when I hire them
• Delivering value to my clients as a Software Architect on their team(s)

Key Clients:

• Hella Infra Market Pvt. Ltd: Manage and support their internal Platform Engineering team. Also playing the role of part-time architect to their Data Platform Team that supports internal data pipelines for ETL, analytics and reporting.
• Astranova Labs: Built their end-to-end infrastructure automation for software delivery on GCP, Kubernetes and Cloudflare.
• Punt Partners: Part of the early team that built the early version of ShelfRadar.ai platform and Tolmol.app product.

Principal Engineer

Content Sourcing Platform: Automated detection of Sales and Use Tax changes across various states within US, reducing time from 3+ weeks of manual effort to a few minutes across many jurisdictions. Architected a No-Code Self-Serve Platform for Tax Researchers within Avalara to automate their workflows to identify important content changes.

Principal Engineer & Mentor

• Data Services Platform: Architected a Data Services Platform (Carol) to run our internal Big Data Workloads. This tool helped orchestrate our internal Spark based platform that ran on EMR clusters created on the fly.
• Product Matching: Consulted for Product Matching team to reduce their workload runtime across 2+ Billion product dataset from multiple days to about 7 - 8 hours. This saved us a lot in Infrastructure costs and also allowed us to run various experiments faster to improve our model metrics: precision and recall faster.
• Data Ingestion: Led the Data Ingestion Team responsible for building tools that ingest all the required data for building our catalogue. To better explain it in numbers: Crawler crawls >20 million webpages a day, Parser process >30 millions webpages a day all in realtime. This team was also responsible for managing all the Master (Raw) Data that flows within our Data pipeline.
• Finder: HTML Archive Storage system for storing over 3 billion HTML documents. It was built using Suuchi, a library that provides set of abstraction for building distributed data systems that I helped develop.
• Infrastructure: During this phase we had to revamp our Devops team internally and as part of the effort, I joined as the first member to help build the team, identify tools, setup automation and processes which after the first year saved over million dollars a year on AWS by intelligently using Spot instances and Auto Scaling for 70% of Production workloads.

Software Engineer

• Data Pipeline: Part of the team that built the first generation of Indix Data Pipeline based on λ-architecture principles. Contributed to writing Map-Reduce jobs, setting up S3 file layouts for faster MR job startup times and monitoring the overall health of the data pipeline.
• Data Ingestion & Data Infrastructure: Responsible for key components like Crawler, Scheduler, Parser, and Master Data Management on S3.

Big Data Intern

• Evaluated AWS for migrating to cloud from custom data centers
• Worked on running R code on Storm