- SDE II at Amazon, working on large-scale compliance infrastructure
- Previously SWE II at Apple: macOS tooling, 5G protocol testing, LLM triage automation - 80% adoption on internal tools, 40% faster incident resolution, $720K annual savings
- Earlier: PayPal full-time; internships at Apple, Tata 1mg, Cisco, Licious, YASH Technologies
- Open-source engineering across RAG and agent infra: retrieval correctness, agent runtime hardening, plugin triggers, sandbox safety, observability, protocol tooling - mostly in RAGFlow, Dify, Langfuse, NDN, Infinity, CustomNerd
- Research with NYU, WSU, UCLA, UBC, Edinburgh - CustomNerd accepted at MDPI Electronics, JobMatch in preparation for JAAMAS
- B.E. CSE, Thapar Institute - CGPA 9.71, top 0.5% merit scholarship
- 1,723 LeetCode (568 easy / 916 medium / 239 hard, rating ~1,693) · 450+ Topmate sessions at 4.3/5
| Focus | Details |
|---|---|
| Building | Large-scale compliance infrastructure |
| Open-source work | Retrieval correctness · agent runtime hardening · plugin triggers · sandbox safety · observability · protocol tooling |
| Exploring | RAGFlow · LangGraph · Dify · Infinity · pydantic-ai · OpenHands · llama_index · fastmcp · together-python · openinference · Phoenix · Ragas · Qdrant |
| Research | CustomNerd framework (NYU, accepted at MDPI Electronics) · JobMatch (JAAMAS, in preparation) · Odyssey / DataFusion cost modelling (Edinburgh) · ScientficPerplexity (UBC) |
Open-source contributions and research directions I'm pushing into over the next 6–12 months.
| Track | Direction |
|---|---|
| RAG infra | infiniflow/ragflow — agent inputs, DeepDoc parsing, clustering, and retrieval-quality edge cases. |
| Agent frameworks | langgenius/dify, dify-sandbox, and dify-official-plugins — sandboxing, plugin auth, and model-provider plumbing. |
| LLM observability | langfuse/langfuse and the neatlogs SDKs — endpoint validation, transport replay guards, SDK init consistency, span routing, and provider isolation. Interested in OTel-native tracing, evals, and feedback loops. |
| Vector & AI-native DBs | qdrant-client, infinity, and ragas — evaluation harnesses for CustomNerd outputs. |
| Protocols & mobility | named-data/ndnd security/CRL work · background in 5G protocol testing. |
- CustomNerd — accepted at MDPI Electronics
- JobMatch — in preparation for JAAMAS
- Odyssey — cost-model work with the University of Edinburgh
- ScientficPerplexity — with UBC
Focused on RAG systems, agent frameworks, observability, protocol tooling, and practical reliability work in open-source codebases.
| Company | Role | Period | |
|---|---|---|---|
| Amazon | Software Development Engineer II | Jul 2026 – Present | |
| Apple | Software Engineer II | Dec 2023 – Jun 2026 | |
| PayPal | Software Engineer | Jul 2023 – Oct 2023 |
Internships
Research
| Work | Venue | Status |
|---|---|---|
| CustomNerd: A Framework and Tool for Fast Deployment of Production-Ready Expertise-Based QA Systems | MDPI Electronics, 2026 | First author, accepted |
| JobMatch: An Agentic Multi-Role Platform for Explainable Job-Candidate Matching | JAAMAS | Co-first author, in preparation |
| InvestorNerd and NewsNerd | NYU, Prof. Dennis Shasha | Contributing author, in preparation |
Core technologies I use day-to-day — not an exhaustive list.
|
Languages & Backend Cloud, Data & AI |
Web & UI Tools & Observability |
|
RAGFlow RAG engine |
Dify sandbox · plugins |
LangGraph Resilient agents |
||
|
Infinity AI-native DB |
ndnd Named-data net |
FastMCP MCP server tooling |
Together Python LLM SDK fixes |
|
Langfuse LLM evals & observability |
Neatlogs Python · Go · TypeScript SDKs |
OpenInference Open LLM tracing |
|
GitHub |
LeetCode |
| School | Degree | Period | Score |
|---|---|---|---|
| Thapar Institute of Engineering and Technology | B.E. Computer Science | 2019 – 2023 | CGPA 9.71 |
| Mount Assisi School | Class XII (ISC) | 2017 – 2019 | 96.25% |
| Mount Assisi School | Class X (ICSE) | 2015 – 2017 | 94.33% |
|
Hobbies Cricket · F1 · Treks · Gym · Cooking · Reading |
Volunteering TICC (Vice Lead Ambassador) · DiiDi · IDIA |





