Ashishkumar Pandey

Solutions Architect & Backend Engineer

Mumbai, India

I build backend systems, streaming pipelines, and OTT platforms on AWS. Currently working on agentic AI on AWS: agent architectures on serverless and Amazon EKS, evaluation, observability, and governance.

Expertise

Backend & APIs

I build REST APIs with sub-30ms p95 latency, handle OAuth 2.0 and OIDC auth flows, and implement idempotency and rate limiting. Mostly with Node.js and TypeScript.

Cloud & Serverless

I work extensively with AWS services like API Gateway, Lambda, DynamoDB, S3, CloudSearch, Kinesis, and MediaConvert. Cost optimization is always on my radar.

Streaming & Media/OTT

I've architected end-to-end OTT platforms with media ingestion, transcoding pipelines, content protection, and Akamai CDN integration. Real-world scale.

Agentic AI on AWS

I'm building agent architectures on AWS serverless and Amazon EKS, with a focus on evaluation and observability, agent authentication/authorisation and logging design, and governance.

IoT & Data Pipelines

I design MQTT and HTTPS ingestion with secure provisioning and multi-tenant isolation, near real-time pipelines on Kinesis/MSK, and time-series and lakehouse storage on Timestream, InfluxDB, S3 and Athena.

Experience

Agentic AI on AWS

Independent

Aug 2026 – Present

Participant in AWS BeSA Cohort 10, "Agentic AI: POC to Production on AWS", an AWS-mentored program on agent architectures, security, observability and governance. Passed AWS Certified AI Practitioner (AIF-C01) plus six certifications across agent design, deployment on AWS serverless and Amazon EKS, evaluation and observability, auth and logging design, and AI governance.

Personal commitments

Career Break

Feb 2026 – Aug 2026

Returned to full-time engineering focus in August 2026.

Solutions Architect

KVell Dynamics

Aug 2024 – Jan 2026

Architected large-scale IoT ingestion for smart meters and sensors over MQTT/HTTPS with device registry, secure provisioning and multi-tenant data isolation. Built near real-time streaming pipelines (IoT Core, Kinesis/MSK, Lambda/Spark) with windowed aggregations, late-event handling and idempotent writes. Designed time-series and lakehouse storage (Timestream/InfluxDB, S3, Glue/Athena), plus analytics, alerting and X.509/TLS and IAM/RBAC security.

Technical Architect & Lead Backend Engineer

Planet Marathi Group

Feb 2021 – Aug 2024

Architected an end-to-end OTT platform across web and mobile, and single-handedly built the backend from concept to production on AWS serverless. Automated media ingestion and transcoding pipelines, integrated Akamai, Razorpay, Twilio/2factor and MediaConvert, and replaced a high-cost auth vendor with in-house OIDC/OAuth 2.0. Achieved sub-30ms p95 APIs and 18+ months of zero downtime. Advised the wider org on engineering practice.

Programmer Analyst

BrandMusiq

Aug 2019 – Jun 2020

Built a React and Redux PWA with offline audio caching in IndexedDB, backed by a Ruby on Rails API for tracks and metadata.

Software Development Engineer

Valuefy Solutions

Sep 2018 – Aug 2019

Shipped OCR validation, face verification/liveness, and secure payment and SMS integrations. Authored build/deploy guidelines and RFP documents, and trained engineers on Git, ES6, NodeJS and Linux.

React Native, Electron, CRM, Rails

Earlier Roles

2015 – 2018

Software Engineer at BrandMusiq: built soniqspaces, a multi-track music player for Android and iOS in React Native, with an Electron desktop version. Lead Developer at Bridge Properties: designed the full stack for the company website, its API and a CRM. Earlier work in Rails, Ionic and hosting support.

Projects

Planet Marathi OTT Platform

Full serverless architecture on AWS. API Gateway and Lambda for compute, DynamoDB for state, Akamai CDN for global distribution. MediaConvert for transcoding. In-house OIDC and OAuth 2.0 auth system.

AWSLambdaDynamoDBAPI GatewayMediaConvertAkamaiOAuth 2.0

IoT Ingestion Platform

MQTT and HTTPS ingestion at scale for smart meters and sensors, with device registry, secure provisioning and multi-tenant data isolation. Streaming pipelines on IoT Core, Kinesis/MSK and Lambda/Spark; time-series and lakehouse storage with Timestream, S3 and Athena; dashboards and alerting with Grafana and QuickSight.

MQTTIoT CoreKinesisMSKLambdaTimestreamGlueAthenaGrafanaQuickSight

BrandMusiq PWA

React and Redux frontend with IndexedDB for offline audio caching. Rails API backend. Works seamlessly online and offline.

ReactReduxIndexedDBRuby on RailsPWA

Skills

Languages & Frameworks

Node.jsTypeScriptJavaScript/ES6Ruby on RailsReactReduxReact NativeElectron

AWS

LambdaAPI GatewayDynamoDBKinesisMediaConvertIoT CoreS3CloudSearchTimestreamGlueAthenaMSK

AI & Agents

Agentic AI on AWS (serverless & EKS)Agent designAgent evaluation & observabilityAI governanceAgent authN/authZ & logging

Domains

Serverless architectureOTT/media pipelinesIoT ingestionStreaming & near real-timeTime-series & lakehouseMulti-tenant data isolationHigh-performance APIs

Integrations

OAuth 2.0OIDCRESTWebhooksRazorpayTwilio2factorAkamai CDN

Ops

p95 performance budgetsLoggingMonitoringAlertingCI/CDBlue/green deploysIncident responseCost governance

Other

MQTTInfluxDBSparkGrafanaQuickSightX.509/TLSIAM/RBAC

Learning

PythonTerraformGo

Certifications

  • AWS Certified AI Practitioner (AIF-C01)

    Amazon Web Services

    Sep 2026

  • AI Governance on AWS

    DataExpert.io

    Sep 2026

  • Agentic AI Evaluation and Observability

    DataExpert.io

    Sep 2026

  • Design Authentication, Authorization and Logging

    DataExpert.io

    Sep 2026

  • Building Agentic AI on Amazon EKS

    DataExpert.io

    Sep 2026

  • Building Agentic AI with AWS Serverless

    DataExpert.io Academy

    Aug 2026

  • Anatomy of an AI Agent

    DataExpert.io Academy

    Aug 2026

Hello World

·2 min read

First post. Setting up the blog, figuring out what goes here.

Get in Touch

Open to interesting backend and architect roles. Let's talk.