MS Artificial Intelligence @ University at Buffalo (Dec 2026)
AI/ML Engineer - I build production LLM systems, RAG pipelines, and multimodal AI applications.

Previously at NxtGen Cloud Technologies, where I shipped AI solutions for enterprise and government clients across multilingual RAG, ASR pipelines, fine-tuning with NVIDIA NeMo, and LLM-powered SaaS products. Before that, built a production RAG pipeline at Harman International (Samsung).

Currently: Data Analyst Intern @ Everly Bly and Co., NYC | Open to full-time AI/ML roles starting Jan 2027.

LLMs & RAG

Vector DBs & Data

Infra & MLOps

RAG-based Socratic tutoring system for Occupational Therapy students. Two-step LLM masking pipeline (Llama 3.1 via Groq) generates guided questions without leaking answers. Multimodal VLM module (Gemma 4) handles anatomy diagram Q&A.
ChromaDB LangChain Llama 3.1 Gemma 4 RAGAS Streamlit GCP
→ RAGAS faithfulness improved 0.70 → 0.80 | 80% anatomical region identification on held-out diagrams

State-level forecasting pipeline for 1–3 month overdose mortality prediction across 54 US jurisdictions. ARIMA vs Prophet benchmarked on 81k+ CDC records. Full pipeline re-implemented on Apache Spark using Medallion architecture.
Apache Spark Databricks ARIMA Prophet K-Means Delta Lake Streamlit
→ Ridge Regression R²=0.997 | K-Means identified 3 distinct state risk profiles

• Agentic AI systems and multi-agent orchestration with LangGraph
• Fine-tuning and RLHF for domain-specific LLMs
• NYC AI/ML community - always up for interesting conversations