AT

Atharv
Tare

Front end, back end, cloud: one engineer, proven at 5M users.

I build the screens people use, the APIs behind them, and the infrastructure that keeps both running on AWS and GCP. Right now I run all of the infrastructure at CogniAI.

Email me

Your visit to this page, traced live

measuring

    Your request stopped at a static file. A real product keeps going, and I've built every hop:

    1. InterfaceReact, Next.js, TypeScript
    2. ServicesNode.js, Go
    3. Events and dataKafka, MongoDB, Postgres
    4. PlatformKubernetes, ArgoCD, Terraform
    5. Cloud and costAWS, GCP

    Interface

    The part people see

    Dashboards, forms and marketing sites in TypeScript and Next.js. For Forethought I built a client's data-entry forms and dashboard from the first sketch to the production deploy.

    This page is hand-written HTML, CSS and JavaScript, with no framework. For you it loaded in a fraction of a second.

    Services

    The APIs behind it

    Node.js and Go services for an app with 5 million users and 400,000 of them active every month.

    • 3× faster main API after reworking caching and MongoDB indexes
    • 65% lower response latency for users

    Events and data

    Work that can't wait for a request

    I moved core workflows onto Kafka, so traffic spikes queue up instead of knocking things over. I also wrote a backup tool that copies MongoDB, Postgres and SQLite to S3 and posts the result in Slack.

    • 50% more throughput through the event pipelines
    • 40% better fault tolerance at peak load

    Platform

    Deploys nobody has to babysit

    GitHub Actions and ArgoCD with blue-green releases, so a bad deploy rolls back in one step. OpenTelemetry and SigNoz wired to Slack, so the team hears about a problem before users do.

    • 2 hours to 10 minutes to stand up a new environment with Terraform
    • Zero downtime on releases across production and staging

    Cloud and cost

    Paying only for what you use

    At CogniAI I'm the only infrastructure engineer. I moved production from AWS to GCP, EKS to GKE and RDS to Cloud SQL, and brought the infrastructure bill down by 30%.

    On one GKE cluster, pods asked for 16 CPU cores and used about 3. I resized all 30 services from real usage, moved them onto smaller autoscaling nodes one group at a time, and nothing went down.

    • $922 to $630 a month for that cluster
    • 53% less CPU reserved for the same traffic

    Where I've worked

    1. 2024 to nowDevOps Engineer, CogniAI
    2. 2023 to 2024Software Engineer, AppyHigh
    3. 2022 to 2023Software Engineer, intern, AppyHigh
    4. 2022 to 2023Full-stack developer, freelance, Forethought
    5. 2021 to 2022Android developer, intern, IBDTD Technologies

    Read what I did in each role

    Not hiring right now?

    Keep my card anyway. When a frontend, backend or infrastructure role opens on your team, I'm one email away.

    Email me LinkedIn GitHub