Software Engineer · Backend & Applied AI

Mohan AmriteshDasari.

I build reliable backend systems and put AI to work inside them — on the Microsoft cloud, and on-device when the data can’t leave.

Backend, platform, and applied-AI engineering — from a Microsoft 365 posture platform and document-AI tools to Snowflake pipelines and fault-tolerant distributed systems.

Get in touchRésumé
years in production
2+
projects shipped
18
Computer Science, UIUC
M.S.
Mohan Amritesh Dasari standing under autumn maple leaves

Currently · SDE @ Princeton IT Services

01About

Systems that survive the real world.

Most of what I build is backend: systems that keep working when things go wrong and, increasingly, the AI features that live inside them. My day-to-day is TypeScript and Python on Azure — Next.js and Flask services, Postgres and Snowflake underneath, Microsoft Graph and Entra ID everywhere, and Docker holding it together.

The roots are in distributed systems — a fault-tolerant ten-node cluster built from scratch at UIUC, then message queues at quick-commerce scale. Since then I’ve shipped a Microsoft 365 posture platform, document-AI tools that run in the cloud and on-device, data pipelines, and security tooling, usually as the engineer who owns the whole stack.

“Ingesting new knowledge and leveling up — one message at a time from the Kafka queue of experience.”

— a personal motto
topic: experienceoffset 0 · lag 45
request
GET /topics/experience/partitions/0/records?offset=0&count=1
Accept: application/vnd.kafka.json.v2+json
X-Consumer-Group: visitors
responsenot sent yet

// 45 records waiting on this partition.
// Send the request to consume the first one.

level 1

// what I focus on

  1. 01

    Platform & backend

    Multi-service systems on Azure: Next.js fronts, Python workers, Postgres as the system of record, and the schedulers, locks, and heartbeats that keep them honest.

  2. 02

    Applied AI

    Document intelligence and language models put to work: cloud extraction pipelines, on-device inference that never leaves the machine, and Copilot integrations.

  3. 03

    Data engineering

    Pipelines that move messy archives into Snowflake, and the reporting built on top of them.

  4. 04

    Identity & security

    Entra ID sign-in and group-based access, Microsoft Graph at scale, email-compromise forensics, and DNS record validation.

02Skills

Tools I build with

The languages, frameworks, and infrastructure behind the work — daily drivers first.

Languages

Frameworks

Data & infrastructure

Cloud & AI

daily drivers×n pick a skill to see the projects built with it

03Experience

From campus to production

Four stops so far — each one a place that changed how I think about building software.

  1. Education2018 — 2022

    VIT Amaravati

    B.Tech in Computer Science — Networking & Cyber Security · GPA 8.9 / 10

    Four years of fundamentals, networks, and security — and the discovery that systems are more interesting when they have to survive the real world.

    • C++
    • Java
    • Networking
    • Security
  2. InternshipJan 2022 — Jul 2022

    Blinkit (previously Grofers)

    Software Development Engineering Intern

    First taste of scale: live order tracking with real-time location plotting, a new message delivery service, and cron automation for rate cards and user onboarding.

    • Python
    • Kafka
    • Redis
    • Celery
  3. Education2022 — 2023

    University of Illinois Urbana-Champaign

    M.S. in Computer Science — Distributed & Cloud Systems · GPA 3.43 / 4.0

    Fault tolerance, failure detectors, and versioned file systems — capped by a 10-node distributed ML cluster, built from scratch, that survived cascading failures.

    • Distributed Systems
    • Cloud
    • PyTorch
    • gRPC
  4. Full-timeFeb 2024 — Present

    Princeton IT Services

    Software Development Engineer

    Shipping for clients and for the company itself: a Microsoft 365 posture platform, document-AI tools that run in the cloud and on-device, enterprise Teams chatbots, data pipelines into Snowflake, and security tooling.

    • Azure
    • TypeScript
    • Python
    • Microsoft Graph
to be continued…

04Projects

Selected work

Four projects that show how I work — platform engineering, applied AI in the cloud and on-device, and what I build after hours.

  1. 01

    M365 posture platform

    Three-service security & operations platform

    Princeton IT Services · product · Feb — Nov 2026

    A Next.js web app, a Python worker, and Postgres — dashboards over Microsoft Graph inventory with live drill-down and revoke, Copilot usage and agent-risk views with quarantine controls, a locked scheduler for ingestion jobs, and signed licensing.

    Inventory, risk, and remediation for a whole tenant

    • Next.js
    • Python
    • PostgreSQL
    • Microsoft Graph
    • Azure
    Full write-up
  2. 02

    On-device intake desk

    Local document AI — nothing leaves the machine

    Princeton IT Services · show-floor demo · Aug — Oct 2026

    Drop a broker packet in and watch a structured submission build itself: a Python sidecar reads geometry, tables, and checkboxes; a local language model fills every field; conflicts become review items with click-to-source provenance.

    A wire log that proves nothing left the machine

    • TypeScript
    • Python
    • Foundry Local
    • Ollama
    • PostgreSQL
    Full write-up
  3. 03

    Business Card Intelligence

    Mobile-first lead capture and follow-up CRM

    Princeton IT Services · product · Apr — Oct 2026

    Photograph a card, get a lead: Document Intelligence extracts it, Azure OpenAI normalizes it, Postgres keeps it, and Graph drafts the follow-up in thread — with two-way Outlook sync, reminders, and a company news feed.

    From a photo to a threaded follow-up

    • Next.js
    • Azure OpenAI
    • Document Intelligence
    • Microsoft Graph
    Full write-up
  4. 04

    Dave, a self-hosted AI agent

    Code-review agent on a local LLM

    Personal project · Oct 2026

    A Discord assistant on a 27B open-weights model served from a DGX Spark with local GPU failover, grown into a GitHub pull request reviewer and repo watcher, with a sandboxed shell whose only way out is a logging proxy.

    Reviews pull requests on its own hardware

    • Python
    • llama.cpp
    • Docker
    • GitHub Apps
    Full write-up

18 projects, grouped by company

Browse the full archive

05Contact

The next chapter is unwritten.

I’m open to new opportunities. Whether you have a role in mind, a question, or an idea worth building — let’s talk.

dmamritesh@gmail.com