Engineer working across AI agents, full stack products, and machine learning. I turn hard, ambiguous problems into things people can actually use, from LLM driven agents to published research.
I work across software development, machine learning, full stack engineering, and distributed systems. Most recently I was an AI Engineering Intern at Postman in Cupertino, building LLM driven agents, deploying production agents, and shipping tools that removed manual work from real workflows.
Before Postman I spent two years at Ericsson R&D delivering more than 100 features for 5G revenue management, plus internships at Fiserv and Samsung R&D. Alongside industry work I have published 10+ research papers across IEEE, AIP, and IGI. I like taking an ambiguous, hard problem and turning it into something people can actually use.
I have published 10+ peer reviewed papers.
Tap a question
The things I build from scratch, from my own AI ventures to a product I shipped inside Postman. This is what I stay up late for.
Teams can build an AI workflow in a week and still not know whether it can be trusted to run a business process. Corda connects to a workflow you already have, benchmarks it against your own historical cases, runs it in shadow beside the human process, then widens its authority one stage at a time as the evidence holds.
An AI recruitment engine that parses job requirements, scores candidates against customizable priorities, and generates LinkedIn x-ray searches. It integrates with the Manatal ATS and exports to CSV, PDF, and JSON.
You build the logic visually in Postman Flows, or just describe it to an LLM, and Portage turns it into a live, running app on whatever backend fits, no plumbing required. Think of it as a forwarding desk for flows: it reads what you built, works out what it needs, generates the files each backend wants, and ships it. Flows builds the logic, Astro runs the agent, and the flow itself never changes, so you can let Portage pick the best destination or point it somewhere new yourself.
A mix of AI agents, multimodal generative systems, biomedical signal processing, and low level systems. From research papers to shipping apps.
An agent that ranks a role's applicants or sources across an entire candidate database. It embeds every candidate into a persistent Qdrant index, retrieves the top matches by similarity, then deep scores them with an LLM using scoring weights that adapt per job description. Email and scheduling happen in the same window.
A clinical web app that analyzes cerebral blood flow physiology in LVAD patients. From raw signals sampled at 125 Hz it computes autoregulation (CA and Mx), cerebrovascular reactivity (CVR), and per beat pulsatility, separating heart driven from pump driven beats. A full Python and Flask port of a MATLAB research tool, with Excel and JSON export.
A multimodal pipeline that drops you into a movie scene. It captions your photo, writes a scene with an LLM, generates the frame with Stable Diffusion, and face swaps you in, all behind a Gradio interface. Built for Penn CIS 5810.
A Streamlit chatbot that answers free form HR questions by pulling live Workday data and applying company policy. An LLM agent selects read only tools, queries over OAuth2 REST, and lets you switch models from the sidebar. Built internally at Postman, so the code stays private.
Proof based hiring. Employers define exactly what evidence they accept for a role, candidates submit real work artifacts, and AI produces a ProofCard, a pass or fail verdict anchored to actual evidence rather than a resume.
A file system built on N-ary trees and hashing for secure, faster backups, with a JSON based hash tree for change detection. Parallelized with OpenMP, it achieved more than 100 percent improvement over traditional systems.
Published in AIP Conf. Proc. 2916 (2023)
A hybrid feature extraction pipeline combining DWT, PCA, GLCM, and OKE on the BRATS 2017 dataset, improving medical image retrieval accuracy by up to 28 percent over baseline models.
A deep learning model that grades diabetic retinopathy from retinal scans using CNNs and an Inception v3 backbone, wrapped in a Flask frontend for easy use by clinicians.
Six sides of how I think and build. Tap any tile and it opens up.
Tuned to what teams are hiring for right now, agentic AI, LLM systems, and production grade full stack and cloud.
More than 10 peer reviewed papers across machine learning, computer vision, healthcare, quantum computing, security, and smart cities.
Philadelphia, PA
Master of Science in Engineering in Computer and Information Science
Expected May 2027
Vellore, India
Bachelor of Technology in Computer Science and Engineering
2019 to 2023