AI Engineer · Senior Software Engineer · Systems Builder

Harshil Gupta

I build AI agents that help SRE teams investigate incidents and operate complex systems.

At Salesforce, I am building Aura, an agentic reliability system using Mastra. My work spans operational tooling, distributed systems, and cloud data platforms.

Selected work

Engineering case studies

SRE / Agentic Systems TypeScript · Mastra · MCP

Aura

An internal Salesforce initiative building AI agents for SRE workflows, bringing operational context and diagnostic tools together to support incident investigation.

SRE agents Mastra Slack MCP
AI / Mobile Product Kotlin · Compose · OpenRouter

LegalBuddy

An Android assistant for grounded legal guidance, with evidence-backed responses and claim verification designed to make answers easier to inspect.

MVVM Room SSE

About

AI x Software Engineering x Cloud Infrastructure

I work at the intersection of AI, software engineering, cloud architecture, and reliability. My background spans AI/ML, backend systems, distributed infrastructure, data platforms, and agentic product development.

I care about building systems that are not only intelligent, but also robust, observable, and dependable under real operational conditions.

What I build

AI Agents Reliability Automation Cloud Systems Data Platforms Developer Tools Intelligent Workflows

Experience

Operating at the edge of AI and production systems

AI Engineer / Senior Software Engineer

Salesforce · Bangalore · Dec 2025 - Present

  • Building Aura, an AI-agent initiative for SRE workflows using the Mastra orchestration framework.
  • Connecting Slack MCP integrations and operational diagnostics to support incident investigation.
  • Developing agentic workflows and production tooling for reliability engineering across distributed systems.

Senior Software Engineer / Software Engineer

NatWest Group · 2023 - 2025

  • Onboarded 5+ data products using OBDEF and AWS Deequ checks to improve data quality across data in transit and at rest.
  • Migrated on-premises Oracle data to S3 with Airflow workflows, reducing storage costs by up to 25%; managed EMR configuration that reduced compute costs by nearly 40%.
  • Built SQL transformations for nested data and Kafka-to-Parquet pipelines on S3; supported DataZone integration, testing, and Terraform-managed AWS infrastructure.

Machine Learning and Operations Intern

Royal Bank of Scotland · Gurugram · June 2022 - Aug 2022

  • Assisted the Data and Analytics team in building an MLOps framework for its new customer conversation intelligence (CCI) model.
  • Developed monitoring scripts to support smooth execution of the CCI model and helped gather label data for model validation.

Research

Applied ML and systems thinking

EEG-based Emotion Recognition

Emotion Recognition through multimodal affective signal analysis

Explored classification systems for valence, arousal, dominance, and liking dimensions using machine learning techniques and signal interpretation pipelines.

Research Python · ML Paper

Skills

Capabilities across AI, systems, and platforms

AI & Agentic Systems

LLMs AI Agents Agentic Workflows MCP Mastra RAG Evaluation AI Automation

Backend & Systems

TypeScript Python REST GraphQL Distributed Systems API Design

Cloud & Infrastructure

AWS GCP S3 EMR Glue Athena Lambda IAM CloudWatch Terraform IaC

Data Engineering

Kafka Airflow MWAA Data Lakes ETL/ELT Data Quality

Development

Git Docker CI/CD Testing Observability

Highlights

Selected accomplishments

GCP AI Skills Hackathon winner
AWS Developer Associate
AWS Cloud Practitioner
Published AI/ML research
AI agent engineering
Cloud and data engineering experience

How I think

I like building systems that...

Automate repetitive engineering work.

Turn operational knowledge into software.

Make complex infrastructure easier to operate.

Use AI where it creates measurable leverage — not just because AI is fashionable.

Contact

Let’s build something useful.

I’m interested in AI systems, developer tooling, reliability engineering, infrastructure, and ambitious technical problems.