M.S. Computer Science
Oregon State University, USA
GPA 3.93 / 4.00
Software Engineer at Microsoft
Hi, I'm Dharun Narayanan
I build as a Software Engineer
I design and ship systems where the engineering is rigorous and the experience is humane. At Microsoft I work on Copilot App Chat in SharePoint, and I build the agent harnesses and loop engineering my team uses to automate day-to-day development, so we can keep pace with how fast AI is changing the way software gets written.
01 About
Engineering that respects the person on the other side
I'm a software engineer at Microsoft. Part of my work is Copilot App Chat in SharePoint. The rest is internal: I build agent harnesses and the loop engineering around them so my team can automate its own day-to-day development, and work out which parts of that workflow are actually worth handing to an agent.
Before this I earned an M.S. in Computer Science at Oregon State University, researching provenance-based intrusion detection. Along the way I interned at Microsoft and Amazon, and spent a year at ORTSOC finding vulnerabilities before someone else did.
The thread through all of it is human-centred design. Good engineering is invisible; bad engineering is somebody's bad day.
02 Education
Academic background
B.Tech Computer Science
Amrita Vishwa Vidyapeetham, India
GPA 3.69 / 4.00
Higher Secondary
Rasi International School, India
89% aggregate
03 Skills
What I work with
Filter by capability rather than reading six lists — the shape matters more than the count.
- Python
- TypeScript
- Java
- JavaScript
- C / C++
- SQL
- Dart
- React
- Node.js
- Flutter
- HTML
- CSS
- REST APIs
- PyTorch
- Deep Learning
- NLP
- Computer Vision
- scikit-learn
- Pandas
- Security Engineering
- Vulnerability Analysis
- Intrusion Detection
- Offensive Security
- Threat Modelling
- AWS
- DynamoDB
- SQS
- Docker
- Ansible
- Firebase
- Human-Centred Design
- UX Research
- Accessibility
- Technical Writing
- Mentoring
Showing all 35 skills.
Python
TypeScript
React
AWS
Docker
PyTorch
Java
Node.js
Security
Git
04 Experience
Where I've built things
Seven engineering roles across industry, research and security operations, from big-tech product teams to a university SOC.
-
Shipping Copilot App Chat in SharePoint. Alongside the product work I build the agent harnesses and loop engineering the team uses to automate its own day-to-day development: the scaffolding that lets an agent plan, act, be evaluated and improve, and the judgement about which work is actually worth handing to one.
-
Researched provenance-based intrusion detection by reconstructing system-level causality graphs to surface intrusions that signature matching misses.
-
Delivered a TypeScript online-evaluation framework for SharePoint Online's autofill-columns feature, using GPT-4.1 to categorise columns at scale, and built the interactive internal dashboards the team used to read user engagement.
-
Built an internal tool on AWS SQS and DynamoDB, surfaced through Amazon's front-end MVC framework, to display verification status for VetDiet pet-prescription orders.
-
Analysed vulnerabilities, hardened security infrastructure and ran assessments for the university's regional security operations centre.
-
Supported instruction, grading and student mentoring across computer science coursework.
-
Researched embedded systems and applied machine learning.
-
Mentored students in technology fundamentals. View certificate
05 Projects
Projects
Span across AI/ML, security and full-stack.
Featured · Human-Centred Design
Beaver Rides
A ground-up redesign of Oregon State's shuttle app. I ran the user research, found that riders' actual question was "will I make it to class", and rebuilt the information architecture around that single question rather than around the transit data model.
UX Research
Prototyping
Usability Testing
Accessibility
Featured · NLP
MELOSCRIBE
Generates music and playable piano notation directly from natural language descriptions. The hard part wasn't generation. It was constraining the output so a human pianist could actually read and play the result.
Python
NLP
Transformers
Music Generation
Featured · Deep Learning
Brain Tumour Segmentation (BraTS)
A ResUNet++ pipeline that segments tumour regions from multi-modal MRI volumes. Built around the reality that in medical imaging a false negative and a false positive are not remotely the same kind of mistake.
PyTorch
ResUNet++
Medical Imaging
Computer Vision
E-Library
Department library system for browsing, borrowing and returning books, with role-based access for staff and students.
Hungrezy
Food ordering service with live cart state, restaurant browsing and order tracking.
ESMS Web
Exam scheduling platform that resolves room, invigilator and timing conflicts before they reach a student.
ESMS Mobile
Flutter companion app giving students their exam schedule and change notifications on the device they actually check.
Code-Mixed Translation
Normalises code-mixed text (Hinglish, Tanglish) into a single target language, the messy reality of how multilingual people actually write.
U-Net Segmentation
Shape-outline segmentation built on the U-Net architecture, used as a controlled testbed for encoder-decoder behaviour.
Forex Predictive Model
Machine learning model for forex price movement, with honest backtesting rather than curve-fitted accuracy claims.
Crop Prediction System
Naive Bayes recommender that suggests suitable crops from soil and climate inputs, designed for low-connectivity use.
Phantom Studios
Freelance marketing site delivered end to end: brief, build and handover to a non-technical owner.
06 Contact
Let's build something
I'm always up for interesting conversations: research collaborations, thorny engineering problems, or just comparing notes on building things people actually like using. If you've got something you think I'd find interesting, I probably will.