Hey, I'm Om. I build systems that scale.
Software engineer working across backend, data, and cloud. I build APIs and real-time services, train ML pipelines on large datasets, and run distributed jobs on cloud clusters. M.S. Computer Science at Illinois Tech, graduating May 2027.
A little about me.
I'm a software engineer who ships. The backend and data infrastructure that has to stay fast and reliable as it grows.
Two production internships and an M.S. in Computer Science in progress at Illinois Tech, 3.5 GPA, graduating May 2027. I move across services, data pipelines, and the cloud they run on, and I own features from first commit to production.
I've shipped real systems: a rule-based configuration engine and real-time GraphQL services at Codage Habitation, and a component library serving hundreds of users at Reality-Craft. On the data side, I build ML pipelines over large biomedical datasets and run distributed Spark and Hadoop jobs on cloud clusters.
My toolkit is Python, TypeScript, and SQL on Docker, CI/CD, and AWS. I care about the parts users never see: clean APIs, solid auth, and infrastructure that holds under load. Right now I'm interviewing for backend, data, and infrastructure engineering roles.
What I actually build.
Four areas I work in day to day. Each one shows up in something I have shipped or researched, not just read about.
APIs and services,
built to last.
GraphQL and REST on Node and TypeScript, real-time features over Socket.IO, and data layers on Prisma, MongoDB, and PostgreSQL. Auth, roles, and the unglamorous logic that keeps a product holding together under real traffic.
Models that earn
their keep.
End-to-end ML in Python, from gradient-boosted rankers (XGBoost, CatBoost, PyTorch) for research to applied LLM work: a fully offline RAG pipeline with local embeddings, vector search over ChromaDB, and answers grounded in your own documents.
Scale it out,
keep it up.
Docker, CI/CD, and secrets done right, deployed on Vercel, Azure, and AWS. I have stood up Hadoop and Spark clusters on Chameleon Cloud and benchmarked distributed jobs at scale.
Safe and steady,
by default.
Role-based access, JWT with secure cookies, encrypted real-time messaging, and validation that actually validates. The invisible work that keeps a system trustworthy when it matters most.
Where I've worked.
Full Stack Web Developer
- Designed an XML-driven configuration engine in Node.js and TypeGraphQL that enforced 100+ rules across product types and cut average configuration time by about 65%, down to under two minutes.
- Built real-time validation over Socket.IO that disabled invalid option combinations the moment users clicked, cutting configuration errors by over 50% with sub-second feedback.
- Shipped production features end to end: role-based access control, server-side PDF generation, and admin-managed email workflows.
- Containerized and deployed with Docker and Vercel on a four-person team, owning backend debugging, performance profiling, and documentation through to client delivery.
Web Developer
- Built 10+ reusable React and TypeScript components for a blog serving 300 to 400 active users.
- Integrated REST APIs to serve dynamic content with category and date-based filtering.
- Fixed layout-breaking bugs and hardened responsiveness across three breakpoints.
- Shipped features through an agile Git workflow: 8+ pull requests with weekly documentation.
Things I've built.
Chat with your docs,
fully offline.
A fully offline retrieval-augmented generation (RAG) system. Upload PDFs, text, or Markdown and ask questions in natural language with nothing leaving the machine: a local LLM and embeddings through Ollama, semantic search over ChromaDB, and grounded answers with source citations in about half a second.
- Python
- FastAPI
- Ollama
- ChromaDB
- Docker
- RAG
Where clients,
lawyers & admins meet.
A real-time legal consultation platform that connects clients, lawyers, and admins. I led the backend on a two-person team, owning about 70% of the codebase: three-role access control, per-session messaging over Socket.IO, and GraphQL resolvers tuned for fast lawyer search.
- Next.js
- Prisma
- MongoDB
- GraphQL
- Apollo
- Socket.IO
A studio site built
for real traffic.
The production website and blog platform for a development studio. Reusable components, a content system that scales as the article library grows, and performance and SEO handled as first-class concerns rather than afterthoughts.
- Next.js
- TypeScript
- Tailwind CSS
- SEO
Teaching models to
find new drugs.
Machine learning research that ranks drug and disease associations across a biomedical dataset of 1.2M+ pairs. I owned the data integration and a 27-feature pipeline, and trained the gradient-boosted rankers, XGBoost and CatBoost, with XGBoost reaching about 95% Top-5 accuracy.
- Python
- PyTorch
- XGBoost
- CatBoost
- scikit-learn
MapReduce vs Spark,
benchmarked properly.
A big-data systems study benchmarking sort at scale. I generated 64 GB of data (~4 billion BLAKE3 records), built shared-memory hash tables in C++ with OpenMP, then ran Hadoop and Spark clusters on Chameleon Cloud: Spark sorted about 2x faster than Hadoop, scaling near-linearly to eight nodes.
- C++
- Java
- Hadoop
- Spark
- OpenMP
- Chameleon Cloud
Retail, but
actually fast.
A responsive storefront for a Canadian retail brand. Clean design, tight SEO, and load times that keep people browsing instead of waiting.
- Next.js
- Tailwind CSS
- SEO
- Responsive Design
Where I studied.
Illinois Institute of Technology
M.S. Computer Science · Chicago, IL
Silver Oak University
B.Tech Computer Engineering · Ahmedabad, India
Let's talk.
I'm interviewing for full-time software engineering roles across backend, data, and cloud, anywhere in the United States. I can contribute from week one. If you're hiring, let's talk.
ompatel.cse@gmail.com→