🎓 Graduate Computer Science @ Arizona State University (MS, 2024–2026)
💡 Full stack engineer with focus in Agentic AI, LLM Optimization, and Fullstack Engineering
🌐 Building production-ready ML systems, inference-optimized agentic workflows, and scalable ETL/ELT pipelines on AWS.

• Python
• C++
• SQL
• TypeScript

• PyTorch
• TensorFlow
• OpenCV
• OpenAI SDK
• LangChain

• Apache Kafka
• Apache Airflow
• Apache Spark
• Apache Iceberg

• Amazon Web Services (AWS)
• Git
• Docker
• Github Actions

• NumPy
• Pandas
• PostgreSQL
• Streamlit

Software Engineer(AI) Intern – Privv (Jan 2026 – May 2026)

• Engineered a bulk invoicing pipeline using AWS Textract for OCR extraction, mapping, and storage automation.
• Architected agentic customized reporting features for project managers, improving decision support and data flow.
• Built modular RESTful microservices and test suites to improve reliability and production stability for AI workflows.

Data Engineer Intern – Alcon (Feb 2024 – Aug 2024)

• Designed ETL pipelines processing 50M+ events/day on AWS (EC2, S3, RDS).
• Improved deployment cycles by 43% with CI/CD workflows.
• Reduced feature retrieval latency by 38% with schema validation & partitioning.

Deep Learning Intern – Bhargava Infotech (Aug 2023 – Feb 2024)

• Built distributed CNN training pipeline on SageMaker, reducing costs by 12%.
• Processed 9GB+ geospatial data in PyTorch, boosting model F1-score by 17%.
• Deployed models with drift detection monitoring.

Contributed a bug fix to DeerFlow's LangGraph integration by resolving a configuration propagation issue affecting modern LangGraph SDK clients.

Highlights

• Identified root cause in gateway configuration handling
• Implemented fix for LangGraph 0.6+ context-based requests
• Added regression coverage to prevent future breakage

PR: bytedance/deer-flow#3735

• Hiiv → Multi Agentic system for sessioned debates(3 mins), leading to better decision making, backed by arguements of different personas.
• Hotel Booking Agent → Multi-channel AI concierge with SMS/web booking, voice STT pipeline, and Python/FastAPI/PostgreSQL backend.
• Mavenly → AI powered CRM intelligence based on knowledge graph structure.
• Efficient LLaMA-2 Optimization → Reduced inference latency (20%) & memory (14%) with GMQA, RoPE & LoRA.
• OpenMonitor → Real-time YOLOv8 surveillance system (🏆 Hackathon runner-up).
• ZeroETL → Open-source PySpark ingestion library for Apache Iceberg with deduplication and snapshot management.

📫 Email · LinkedIn · GitHub