Senior Software Engineer with 3+ years building production-grade distributed systems, cloud-scale data pipelines, and AI-powered automation. Currently at Rokt, where I optimized billion-scale Spark workloads to deliver a 4× runtime reduction and built end-to-end observability infrastructure for mission-critical pipelines.

I work across the full stack — Spark/Scala data engineering, AWS infrastructure, agentic LLM systems, and backend services — focused on systems that are fast, observable, and built to scale.

mindmap
  root((Engineering Stack))
    Cloud Scale Systems
      Spark/Scala Pipelines
      Apache Iceberg
      Billion-Row Processing
      Memory-Safe DataFrame Design
    Production Infrastructure
      AWS EKS/ECS/EC2
      Kubernetes Orchestration
      Terraform IaC
      Datadog Observability
    AI & Automation
      LLM Orchestration
      Agentic Workflows
      RAG Pipelines
      Multi-Agent Systems
    Backend Engineering
      Python/Java/C++
      REST/GraphQL APIs
      PostgreSQL/MongoDB
      Redis/Kafka Streams
    DevOps & Reliability
      CI/CD Pipelines
      Docker Containerization
      Monitoring & Alerting
      Live-Site Triage

timeline
    title Engineering Journey
    section Foundation
    2022 : Software Engineer
         : Motorola Solutions
         : iOS infrastructure
         : Real-time media systems
    section Research & Innovation
    2021 : Research Intern
         : Indian Institute of Science
         : HMM-DNN ASR model
         : Bill & Melinda Gates funding
    section Growth & Impact
    2024 : Software Engineer
         : Nix Inc
         : Document processing infrastructure
         : AWS cloud automation
    2024 : AI Engineer
         : AI Growth Agent
         : SEO automation pipeline
         : 3x keyword coverage increase
    section Data & AI Engineering
    2025 : Software Engineer
         : Rokt
         : 4x Spark optimization
         : End-to-end observability
         : Billion-scale processing

Mission-Critical Performance Optimization

• Delivered 4× runtime reduction on billion-scale identity graph statistics pipeline by optimizing Spark/Scala execution strategy (4h → 1h), saving $50K+ annually in compute costs
• Architected end-to-end observability system computing 10+ statistical metrics (percentiles, distributions, time-window trends) with Apache Iceberg persistence for reliable downstream analytics
• Provisioned production-grade monitoring using Datadog dashboards/monitors + Terraform IaC, launching alerts for runtime regression, failure rate, executor OOMs, shuffle spill, and data freshness across 10+ mission-critical pipelines
• Pioneered LLM-driven automation by prototyping an agentic content-generation workflow in n8n, cutting drafting turnaround from multi-hour manual effort to sub-10-minute automated runs
• Enhanced backend scalability for services handling large volumes of unstructured document inputs, laying the foundation for search, document understanding, and applied AI workflows

Technologies: Spark, Scala, Apache Iceberg, Datadog, Terraform, LLM orchestration, n8n

AI-Powered Content & SEO Automation

• Increased targeted keyword coverage by 3× through context-aware keyword research logic mining search-volume, intent, and competitor gaps
• Achieved 100% structured-data compliance by generating SEO assets (JSON-LD schema, meta titles, slugs, alt text) directly in the flow, qualifying all articles for Google rich-result snippets
• Cut mean time to recovery by 50%+ through robust monitoring & failure logging that streams real-time pipeline errors to Slack
• Built cloud-based document ingestion infrastructure for preprocessing, parsing, and workflow orchestration, improving reliability and turnaround time of downstream processing
• Developed operational tooling to inspect unstructured-document processing outputs, validate extracted content/metadata, and accelerate iteration on retrieval-oriented workflows

Technologies: LLM pipelines, JSON-LD, Slack integration, cloud document processing

Cloud Infrastructure & Document Processing Platform

• Built production document ingestion infrastructure supporting preprocessing, parsing, and workflow orchestration for cloud-based intelligent document features, improving turnaround time and reliability
• Developed visualization and operational tooling enabling teams to inspect document-processing outputs, monitor system behavior, and rapidly prototype workflow improvements
• Improved backend service reliability and scalability for handling unstructured document inputs, building foundations for applied AI and data-driven product workflows
• Automated CI/CD workflows with Docker, Kubernetes, and GitHub Actions, increasing deployment speed and reducing errors while enabling rapid, high-quality feature releases in AWS production environments
• Integrated third-party APIs and built robust REST endpoints, expanding Nix's platform ecosystem, accelerating strategic partner integrations, and enabling innovative document workflow enhancements

Technologies: Docker, Kubernetes, GitHub Actions, AWS (EC2, EKS, S3, RDS), REST APIs

Mobile Infrastructure & Real-Time Systems

• Improved application infrastructure supporting real-time audio, camera, and device-signal inputs, enhancing stability and performance for multimodal, latency-sensitive mobile workflows
• Optimized memory usage and resource management for media-heavy iOS features, enabling more reliable performance under demanding real-world operating conditions
• Built debugging and validation workflows accelerating iteration on complex communication features involving streaming, device state, and secure data handling
• Coordinated with QA and release teams on test planning and execution, contributing to a 20% reduction in post-release bugs and consistently improving software quality across multiple development cycles
• Enhanced iOS app performance for users in demanding public safety environments and extended operational deployments

Technologies: iOS (Swift, Objective-C), C++, real-time media processing

Machine Learning Research & Speech Recognition

• Designed and trained hybrid HMM-DNN ASR model for Kannada language from scratch, significantly advancing speech tech for low-resource regional dialects
• Led data validation and preprocessing across 10,000+ samples to ensure linguistic consistency, completeness, and formatting integrity for high-quality model training
• Contributed to securing research funding from the Bill & Melinda Gates Foundation by demonstrating the project's potential impact on inclusive voice-based technologies for underserved populations

Technologies: HMM-DNN, speech recognition, Python, data preprocessing

GPU-Accelerated Physics Simulation

• Implemented high-performance Smoothed Particle Hydrodynamics (SPH) simulation in C++ using GPU compute shaders and OpenGL, enabling efficient modeling of fluid dynamics in real time
• Achieved rendering of 200,000+ particles at 30 FPS within a custom graphics pipeline, showcasing GPU-accelerated computation for resource-constrained environments

Technologies: C++, GPU compute shaders, OpenGL, physics simulation

Multi-agent LLM system for threat detection and response automation

What it is: Built production-grade SOC platform combining ML intrusion detection, LLM alert triage, RAG over threat intel, and attack-campaign simulation with response orchestration.

Built:

• Architected multi-agent framework orchestrating specialist AI agents for routing, retrieval, fact-checking, and safety validation
• Integrated Wazuh SIEM with custom webhook receiver and alert enrichment pipeline
• Deployed Prometheus + Grafana monitoring stack for real-time operational visibility
• Implemented D3FEND-mapped countermeasures with approval tiers and verification loops

Technologies: Python, LangChain, Claude Sonnet 4.6, Wazuh, Prometheus, Grafana, Docker, PostgreSQL

Impact: Self-correcting threat detection system with built-in fact-checking and graduated autonomy controls for security automation.

Self-correcting retrieval-augmented generation with hybrid knowledge sources

What it is: Production-grade RAG pipeline with intelligent routing, query reformulation, web search fallback, and integrated fact-checking