Software Engineer | AI/GenAI Systems | Distributed Backends | Published NLP Researcher

Building AI systems that ship to production — agentic backends, RAG pipelines, and distributed microservices at scale. Currently at PNC Financial Services delivering GenAI solutions for banking operations. Published NLP researcher (RANLP 2025). MS Computer Science, CSUDH.

Full-stack AI engineering: from LLM integrations and agentic orchestration through distributed backend systems and cloud infrastructure. Three areas:

1. Agentic systems — multi-agent orchestration with LangGraph, long-term RAG memory, real-time trace streaming, tool use, and circuit-breaker reliability patterns. Systems that complete complex tasks autonomously and don't fall over under load.

2. Production backends — Spring Boot and FastAPI microservices with JWT/RBAC security, async processing, Kafka pipelines, and Redis caching. Built for observability, incident response, and zero-downtime deployment.

3. AI/LLM integrations — OpenAI API, LangChain, vector databases (Pinecone, ChromaDB, FAISS), embedding-based semantic search, and domain-specific fine-tuning of transformer models. Applied in banking, finance, and clinical NLP.

Python · FastAPI · LangGraph · OpenAI API · Pinecone · Redis · PostgreSQL · React · Docker

6 specialized agents coordinated by a LangGraph orchestrator. Long-term RAG memory via Pinecone. Streaming FastAPI layer with real-time agent trace endpoints and a React dashboard for live step visualization. Redis caching and circuit-breaker retry logic. 91% task completion rate across 1,200+ test runs. 87% automated test coverage.

React 19 · Flask · OpenAI API · LangChain · Scikit-learn · Firebase · SQLAlchemy · PDFPlumber

Full-stack platform with AI-driven spending analysis, bill tracking, goal monitoring, and personalized recommendations. PDF bank statement ingestion that auto-categorizes 95%+ of transactions across 8 major bank formats. Scikit-learn models for spending prediction with 83% accuracy across 15+ categories.

Python · LangChain · OpenAI API · ChromaDB · FastAPI

RAG pipeline over ChromaDB for automated legal document parsing, clause extraction, and contextual enterprise contract search. Conversational query handling over compliance-sensitive documents.

Advancing Clinical Translation in Nepali through Fine-Tuned Multilingual Models RANLP 2025 · Peer-Reviewed · Co-Author

Fine-tuning of NMT models for low-resource clinical NLP in Nepali. Domain adaptation with limited parallel bilingual data. Demonstrates expertise in transformer-based language models, multilingual NLP, and domain-specific LLM adaptation.

Languages: Python, Java, JavaScript, SQL

AI / LLM: OpenAI API, LangChain, LangGraph, RAG, Agentic Workflows, Pinecone, FAISS, ChromaDB, Embeddings, Transformer Fine-Tuning, Scikit-learn

Backend: Spring Boot, FastAPI, Flask, Node.js, gRPC, Microservices, Kafka, Redis

Cloud & DevOps: AWS (EC2, S3, RDS), Docker, Kubernetes, GitHub Actions, Jenkins

Databases: PostgreSQL, MongoDB, MySQL, Oracle SQL

Security: JWT, OAuth 2.0, RBAC, Firebase Auth