Arnav Rajesh - SDE / AI Engineer

Software & AI engineer building agentic systems, backend platforms, and product-grade AI workflows.

01 - About

I'm a Software Engineer with ~3 years of production experience across backend platforms, developer infrastructure, and applied AI workflows. I'm especially interested in roles where technical depth, product judgment, and emerging AI capabilities come together to build tools teams can rely on.

02 - Experience

Where I've built

May 2026 - Aug 2026
IMAI Studio
AI Engineer, Consultant
AI engineering on an all-in-one platform taking products from idea to imagery, video, 3D assets, and manufacturing specs.
- Shipped a production avatar creation workflow, an interactive builder driving Multimodal pipelines for previews and final generations.
- Built a Real-time Voice Agent on that calls tools for live generation status and delegates work to specialized sub-agents.
- Collaborated on a multi-agent orchestration framework around a Virtual File System (VFS) for task delegation, failure recovery, and multi-step generation and research workflows.
- Implemented multi-provider model fallbacks for reliability, plus skill-based workflows for data extraction, image embeddings, and semantic search.
- multimodal generation
- real-time voice AI
- multi-agent orchestration
Jan 2024 - May 2026
PhonePe
Software Engineer, Backend
Owned internal platforms used across the organization, spanning AI incident response, secure database operations, SSO, and certificate analytics.
- Architected Cortex, an AI incident-response system serving as first responder in on-call Slack channels, with Agentic RAG workflows, MCP integrations into live internal tools, and multimodal context so engineers can just paste a screenshot.
- Led DaQuery from inception to deployment: RBAC approval flows, a form-based SQL builder, and Celery/RabbitMQ distributed execution across database fleets with reusable QuerySets, cutting audit effort by up to 80%.
- Built certificate analytics using Python and Elasticsearch covering 30,000+ servers, reaching 100% adoption for certificate management across the company.
- Enhanced and maintained Tools Identity, the org-wide SSO platform on OAuth 2.0 with RBAC and team-level permissions; mentored incoming interns.
- agentic RAG
- distributed execution
- OAuth 2.0 + RBAC
Jun 2023 - Nov 2023
Indian Institute of Science
Research Intern, AI
Collaborated on Ocularone, a drone-based assistive navigation system for visually impaired users; work that fed into the Ocularone-Bench paper.
- Built the real-time inference pipeline: a Python WebSocket server streaming live drone video through YOLOv4 person detection into the drone's control loop.
- Set up low-latency bidirectional communication between the Android controller app and the processing server.
- Collected and annotated the person-detection dataset behind the navigation workflow.
- Python
- real-time video
- WebSockets
Dec 2022 - Jun 2023
Tech Machinery Labs
Frontend Intern, Mobile
Mobile frontend internship building Flutter flows for cattle donation and detector-backed identification, with REST API integration into the backend.
- Built Flutter workflows for cattle donation and image-based identification, turning field intake requirements into mobile-first screens.
- Integrated REST APIs that connected the mobile app to a server-side detector for cattle identification requests and results.
- Implemented form, loading, and response states so donation details, image submission, and identification feedback moved through a clean app flow.
- Flutter
- REST APIs
- mobile workflows

03 - Projects

Selected
work

Five builds across agents, NLP pipelines, privacy tooling, and mobile products.

0105

Makima

model + tool runtime

An AI agent framework that dynamically orchestrated multiple LLMs and external tools through function calling, built before MCP existed. Every user-created agent got its own vector-based knowledge base: a dedicated retrieval source that kept responses grounded and specialized.

Python
LLMs
REST APIs
Vector DBs
Tooling

NERD

question generation + answer scoring

A full-stack NLP platform for test taking and evaluation: retrieval-based question paper generation grounded in uploaded study material, and an automated grading pipeline combining BERT-variant features with an XGBoost regressor, scoring descriptive answers to a 0.95 MAE.

BERT
XGBoost
NLP
Retrieval

ArkhamRush

private paste flow for AI tools

A macOS privacy tool for the age of AI: regular expressions and Microsoft's Presidio detect and obfuscate sensitive data on the clipboard before it ever reaches a model, because pasted context shouldn't leak PII.

macOS
Presidio
Regex
Privacy

RIT Connect

student-facing mobile app

A Flutter app adopted by 1,000+ students at Ramaiah Institute of Technology, centralizing access to academic information and educational resources through interfaces students actually enjoyed using.

Flutter
Mobile
Product
UX

Sahasrara

memory game prototype

Winner of Smart India Hackathon 2022: a Flutter app of memory-based games designed to fight early-onset dementia, with a Firebase leaderboard adding friendly competition to keep patients engaged.

Flutter
Firebase
Games
Health-tech

04 - Contact

From abstract ideasto refined systems

Where I've built

May 2026 - Aug 2026

IMAI Studio

AI Engineer, Consultant

AI engineering on an all-in-one platform taking products from idea to imagery, video, 3D assets, and manufacturing specs.

Shipped a production avatar creation workflow, an interactive builder driving Multimodal pipelines for previews and final generations.
Built a Real-time Voice Agent on that calls tools for live generation status and delegates work to specialized sub-agents.
Collaborated on a multi-agent orchestration framework around a Virtual File System (VFS) for task delegation, failure recovery, and multi-step generation and research workflows.
Implemented multi-provider model fallbacks for reliability, plus skill-based workflows for data extraction, image embeddings, and semantic search.

multimodal generation
real-time voice AI
multi-agent orchestration

IMAI Studio generation platform interface

Jan 2024 - May 2026

PhonePe

Software Engineer, Backend

Owned internal platforms used across the organization, spanning AI incident response, secure database operations, SSO, and certificate analytics.

Architected Cortex, an AI incident-response system serving as first responder in on-call Slack channels, with Agentic RAG workflows, MCP integrations into live internal tools, and multimodal context so engineers can just paste a screenshot.
Led DaQuery from inception to deployment: RBAC approval flows, a form-based SQL builder, and Celery/RabbitMQ distributed execution across database fleets with reusable QuerySets, cutting audit effort by up to 80%.
Built certificate analytics using Python and Elasticsearch covering 30,000+ servers, reaching 100% adoption for certificate management across the company.
Enhanced and maintained Tools Identity, the org-wide SSO platform on OAuth 2.0 with RBAC and team-level permissions; mentored incoming interns.

agentic RAG
distributed execution
OAuth 2.0 + RBAC

Receiving the PhonePe capture-the-flag award

Jun 2023 - Nov 2023

Indian Institute of Science

Research Intern, AI

Collaborated on Ocularone, a drone-based assistive navigation system for visually impaired users; work that fed into the Ocularone-Bench paper.

Built the real-time inference pipeline: a Python WebSocket server streaming live drone video through YOLOv4 person detection into the drone's control loop.
Set up low-latency bidirectional communication between the Android controller app and the processing server.
Collected and annotated the person-detection dataset behind the navigation workflow.

Python
real-time video
WebSockets

Dec 2022 - Jun 2023

Tech Machinery Labs

Frontend Intern, Mobile

Mobile frontend internship building Flutter flows for cattle donation and detector-backed identification, with REST API integration into the backend.

Built Flutter workflows for cattle donation and image-based identification, turning field intake requirements into mobile-first screens.
Integrated REST APIs that connected the mobile app to a server-side detector for cattle identification requests and results.
Implemented form, loading, and response states so donation details, image submission, and identification feedback moved through a clean app flow.

Flutter
REST APIs
mobile workflows

ARNAVRAJESH

Where I've built

IMAI Studio

PhonePe

Indian Institute of Science

Tech Machinery Labs