Where I've worked.
Recurly
Software Engineer Intern
Built internal agent infrastructure for Recurly's engineering and go-to-market teams, from the platform that turns product requirements into merged pull requests to the tooling that researches inbound prospects.
- Built the core PDLC automation platform using specialized Claude agents (planner, decomposer, a Sonnet/Opus coder pair, and a PR responder) deployed as isolated Kubernetes Jobs via the Agent SDK. Drives PRD to merged PRs through five human checkpoints, accelerating shipping about 3x.
- Shipped an AI SDR Slack bot (async Python, 3-tier Claude routing) querying BigQuery, 6sense, ZoomInfo, and SFDC to research, qualify, and audit account and call history. Returns source-cited briefs and outreach drafts in about 25 seconds.
- Automated a Prospect Report Card pipeline (n8n to FastAPI on Cloud Run to Claude Opus) that turns an inbound marketing form-fill into a personalized PDF analysis of the prospect's subscription business and emails it back.
- Claude Agent SDK
- Kubernetes
- Python
- FastAPI
- BigQuery
QualGent
Software Engineer Intern
Reported to the CTO on App Crawler and the QualGent AI Assistant, taking agent infrastructure from prototype into production inside a YC X25 company.
- Architected App Crawler, a GKE-distributed Python system indexing Android apps into an org-wide knowledge base, using a GPT-4o DFS agent with Set-of-Marks and uiautomator2 to traverse UIs via ADB and populate a Vertex AI RAG corpus.
- Engineered the event-driven backend that runs it hands-off: a Supabase-queue watcher spawning per-app Kubernetes Jobs, an AAB to APK converter, and remote emulator leasing from a managed GCE fleet, holding a sub-1% failure rate at scale via self-healing CronJobs.
- Built the flagship QualGent AI Assistant, an enterprise QA copilot: a Gemini 2.5 Pro orchestrator on Google ADK routing across 45+ tools and sub-agents (RAG, MCP Postgres/pgvector Toolbox, Jira and Linear via OAuth), deployed on Vertex AI Agent Engine through gated Cloud Build.
- Google ADK
- GKE
- Vertex AI
- Python
- RAG
Purdue University
Software Engineer, AI Research
Applied ML research where the product constraints matter as much as the model: interpretability, privacy, and whether a clinician would actually trust the output.
- Built CLIP-H for clinical hypothesis verification on MIMIC-IV using Top-K sparse autoencoders and a GPT/Claude ensemble. Validated against a synthetic oracle at 0.844 AUROC, certifying 14 hypotheses for a NeurIPS submission targeted for September 2026 with Purdue and Harvard Business School faculty.
- Developed a Flutter and TypeScript clinical assistant tested across Indiana hospitals that cuts nurses' documentation overhead by about 40%, running self-hosted LLaMA 3.2 and an on-device, HIPAA-compliant speech-to-text pipeline.
- Presented the work at the Purdue Spring Research Conference.
- Research
- MIMIC-IV
- Interpretability
- LLaMA
- Flutter
Microsoft
Software Engineer, Microsoft Research Collaboration
Owned the LLM pipeline and the distributed infrastructure under it, turning raw social chatter about Minecraft into insights that reached Microsoft product and marketing leadership.
- Owned a Python LLM sentiment pipeline (LLaMA 4) that processed millions of social posts across Minecraft and A Minecraft Movie.
- Designed the distributed Azure Databricks and Spark workflows underneath it, with SQL-optimized schemas across ingestion, inference, evaluation, and monitoring under strict cost and latency SLAs.
- Cut compute spend by about 25% with no loss in accuracy.
- LLaMA
- Spark
- Databricks
- Azure
- Evaluation
Awards, rankings, and a near miss.
Y Combinator S26, top 10% of applicants
YC told our Checkpoint team that our Summer 2026 application ranked in the top 10% of the applicant pool. We didn't get an interview.
NeurIPS submission in preparation on clinical hypothesis verification
Certified 14 hypotheses with CLIP-H at 0.844 AUROC. Being prepared for a NeurIPS submission targeted for September 2026, with Purdue and Harvard Business School faculty.
Mary-Ann Neel Computer Science Scholar
Awarded to a top Purdue CS student for academic and technical excellence.
Discovery Park Research Scholar, 3 times
Recognized three times for high-impact interdisciplinary research and engineering.
Dean's List and Semester Honors, 6 times
Six semesters at Purdue, carrying a 3.90 / 4.00 GPA in CS and AI.
KVPY, All India Rank 1638 (top 1%)
Nationally ranked among 150,000+ candidates for exceptional scientific aptitude.
Teaching assistant and workshop lead
TA for CS 471 (AI) and TDM 101, guiding 300+ students. Led weekly Machine Learning @ Purdue workshops mentoring 200+ more on building and deploying production AI systems.